-
Crypto AG: How the CIA and BND Sold Rigged Encryption to 120 Countries for Decades
In 1970, the Central Intelligence Agency and West Germany’s Bundesnachrichtendienst paid $5.75 million for a Swiss encryption company called Crypto AG. They didn’t announce the purchase. They didn’t change the branding. They didn’t replace the employees. They installed one or two people at the executive level who knew the truth, kept the rest of the workforce in the dark, and for the next 48 years sold encryption machines to more than 120 governments worldwide — machines that the CIA and NSA had rigged so that every message encrypted on them could be read by American and German intelligence as easily as plaintext. The governments of Iran, Egypt, Pakistan, Saudi Arabia, Italy, Argentina, India, the Vatican, and dozens of others paid good money for equipment they believed was protecting their most sensitive diplomatic and military communications. It was doing the opposite. A CIA internal history, leaked in 2020, called the operation “the intelligence coup of the century.” That’s not journalistic hyperbole. That’s the agency’s own classified assessment of its own program.
The Hagelin relationship
The story starts before the CIA owned the company. Boris Hagelin, a Swedish inventor, founded Crypto AG in 1952 after building the M-209 cipher machine that the U.S. military used extensively during World War II. Hagelin relocated to Switzerland and built a business selling encryption equipment to governments worldwide, leveraging Swiss neutrality as a brand asset — a company based in a neutral country, manufacturing security products, seemed inherently trustworthy. By the early 1950s, Hagelin had entered an informal arrangement with William Friedman, the NSA cryptologist widely regarded as the father of American codebreaking. The “gentlemen’s understanding” was straightforward: Hagelin would sell his most capable machines to countries approved by the U.S., and weaker, breakable versions to everyone else. The arrangement was unofficial, personal, and — critically — it worked. Correspondence between Friedman and Hagelin, declassified in 2015, documented the relationship in detail.
By the late 1960s, Hagelin was aging and the informal arrangement was becoming untenable. When French and West German intelligence approached Hagelin in 1967 to propose their own partnership, Hagelin reported the approach to his CIA handlers. The agency decided it was time to buy the company outright. They partnered with the BND, and in June 1970 the purchase was completed. Crypto AG was given the internal codename “Minerva.” The operation was initially called “Thesaurus,” later renamed “Rubicon.” Hagelin’s son, Boris Jr., who had been the company’s sales manager for the Americas, died in a car accident the same year. His father investigated and did not believe it was an accident.
How the rigging worked
The manipulation was elegant rather than crude. The CIA and NSA didn’t install obvious backdoors or program the machines to dump their encryption keys. They weakened the algorithms — specifically, they rigged the keystream generators so that the output, while appearing random to the user, contained mathematical structures that the NSA could exploit to recover the plaintext. To anyone without knowledge of the specific weakness, the encryption looked secure. To the NSA, it was transparent. As the technology evolved from mechanical cipher machines to electronic systems to software, the rigging evolved with it. NSA cryptologists and CIA engineers worked with a small number of witting Crypto AG technical staff to design each new generation of products with weaknesses that were invisible to the company’s own unwitting engineers and to every customer who tested the equipment.
Siemens, the German electronics conglomerate, manufactured teleprinters for Crypto AG, provided management personnel for 20 years, and held a five percent share of the profits. Siemens engineers helped develop the encryption equipment. The Maximator alliance — a second Western signals intelligence partnership comprising Denmark, France, Germany, Sweden, and the Netherlands, operating parallel to the Five Eyes — was also read into the vulnerabilities and exploited them for their own intelligence collection. The circle of governments benefiting from Crypto AG’s compromised machines was wider than the CIA and BND alone.
What it produced
The intelligence yield was staggering across decades of global events. During the 1978 Camp David negotiations between Egypt and Israel, the NSA read every communication between President Sadat and his advisors in Cairo — because Egypt was a major Crypto AG customer. During the 1979 Iran hostage crisis, Iranian communications were intercepted in real time. In 1982, the British government received intelligence during the Falklands War because Argentina’s military encrypted its communications on Crypto AG equipment. In 1986, intercepted Libyan diplomatic traffic between Tripoli and the Libyan embassy in East Berlin provided the evidence President Reagan cited when he ordered the bombing of Tripoli and Benghazi in retaliation for the West Berlin discotheque bombing — and Reagan’s public statement about the intercept nearly blew the entire operation, because Libya and every other Crypto AG customer suddenly had a reason to wonder how the Americans were reading their communications.
By 1988, the CIA and BND were decrypting approximately 19,000 Iranian messages annually — 80 to 90 percent of Iran’s total encrypted traffic. The operation provided intelligence on the South American Operation Condor dictatorships — Chile, Argentina, Bolivia, Paraguay, Uruguay, and Brazil — as they coordinated cross-border campaigns of imprisonment, torture, and extrajudicial killing. The Condor nations used Crypto AG equipment to coordinate their operations. American and German intelligence read the traffic. They knew what was happening. The CIA and BND documents, as the Washington Post reported, “largely avoid more unsettling questions, including what the United States knew — and what it did or didn’t do — about countries that used Crypto machines while engaged in assassination plots, ethnic cleansing campaigns and human rights abuses.”
How it almost fell apart — and didn’t
The operation survived repeated near-exposures across five decades, which is arguably more remarkable than the operation itself. Reagan’s 1986 public reference to Libyan intercepts was the first serious scare. The 1991 assassination of former Iranian Prime Minister Shapour Bakhtiar produced another: Iranian intelligence transmitted a coded message to Iranian embassies the day before Bakhtiar’s body was discovered, and the speed of Western intelligence’s response raised suspicions about how the intercept was obtained.
The most dramatic exposure came in 1992, when Hans Bühler, a Swiss Crypto AG salesman, was arrested in Iran on espionage charges. Bühler had no idea he was selling rigged equipment — he was a genuine salesman who believed in his company’s products. Iran detained him for nine and a half months. Crypto AG paid approximately $1 million in bail for his release. When Bühler returned to Switzerland, he started talking to journalists. Another former Crypto AG engineer who had independently suspected the company was controlled by Western intelligence also went public. The media coverage was extensive. Bühler was fired. But the operation survived. The BND, rattled by the exposure risk, sold its stake to the CIA in 1993 or 1994 for $17 million. The CIA kept going alone. For another 24 years.
Why did it survive? An academic study in Intelligence and National Security identified three factors: geopolitical pressures on target countries that limited their alternatives, the target governments’ limited technical resources for independently verifying encryption security, and individual operational brilliance by CIA-BND agents inside Crypto AG who managed each crisis without the operation collapsing. The simplest factor was the most powerful — there weren’t many alternatives. If you were a mid-sized government in the 1980s and you needed encryption equipment, your options were American, Soviet, or Swiss. The Swiss option looked neutral. It wasn’t.
What it means
The CIA sold Crypto AG’s remaining assets in 2018. The Swiss company was split into CyOne (domestic Swiss sales) and Crypto International AG (international sales under new ownership). The operation formally ended after 48 years of continuous signals intelligence collection from more than 120 governments. But the structural lesson is the one that connects Crypto AG to every other lecture in the Shadowcraft course: the most effective covert operation isn’t one that steals secrets. It’s one that sells the target the tool they’ll use to betray themselves — and charges them for the privilege.
The parallel to modern debates about encryption backdoors, tech company cooperation with intelligence agencies, and the post-Snowden landscape is obvious and uncomfortable. As Warwick University researchers noted after the 2020 revelations: “Long before Edward Snowden released documents of modern firms colluding with intelligence agencies, we can see evidence for significant cases in the past. It certainly is not a recent phenomenon and leads us to ask just how many firms had been working directly with intelligence agencies.” The question the Crypto AG story poses isn’t whether intelligence agencies compromise commercial encryption. It’s how many current products carry weaknesses that will take another 48 years to discover. We cover Operation Rubicon alongside BCCI’s financial architecture, the United Front Work Department’s influence networks, Wagner Group‘s mercenary-propaganda fusion, and the shell company structures that make all of it possible across our Shadowcraft course — 24 lectures on the invisible institutions that shaped the modern world from behind the paperwork.
-
The United Front Work Department: How China Runs the World’s Largest Influence Operation
In September 2024, a former aide to the governor of New York was arrested for allegedly acting as an unregistered agent of the Chinese government. The same month, a Chinese democracy activist living in New York was arrested and accused of spying for Beijing. A month earlier, a historian in the same city was convicted of being an agent for China’s intelligence services. Three separate cases, three separate individuals, one city, one operational playbook — and the organization coordinating the playbook has been running continuously since 1942, reports directly to the Chinese Communist Party’s Central Committee, and as of a February 2026 study by the Jamestown Foundation, operates through more than 2,000 linked organizations across the United States, Canada, the United Kingdom, and Germany alone. The organization is the United Front Work Department, and most people outside of intelligence and China-studies circles have never heard of it.
What the United Front actually is
The UFWD is one of six main departments of the Chinese Communist Party. It is not an intelligence agency in the way the CIA or MI6 are — it doesn’t run agents collecting classified information. It is something more structurally ambitious than that. The UFWD’s mission is to identify, cultivate, and manage relationships with every group and individual outside the CCP that could be useful to the Party’s interests — ethnic minorities, religious organizations, private entrepreneurs, overseas Chinese communities, foreign politicians, academics, business leaders, social media influencers, and the eight minor political parties legally permitted to exist inside China. Mao Zedong described the UFWD as one of the Party’s “three magic weapons” alongside the People’s Liberation Army and the Party itself. Xi Jinping repeated that description in 2017. He wasn’t being nostalgic.
The best one-sentence summary came from Representative Mike Gallagher, former chair of the House Select Committee on the CCP: the United Front’s operational principle is “making idiots useful” — co-opting any individual or organization to advance the Party’s goals. The strategy operates through three main channels: silencing criticism of the regime globally, promoting Beijing’s preferred narratives abroad, and manipulating foreign institutions through clandestine and often illegal operations. Stanford University’s Internet Observatory and the Hoover Institution described the United Front as cultivating pro-Beijing perspectives by rewarding those it deems friendly with accolades and lucrative opportunities, while orchestrating social and economic pressure against critics — pressure that is “often intense but indirect, and clear attribution is therefore difficult.”
The difficulty of attribution is the feature, not the bug. The UFWD operates through quasi-official organizations, civic groups, cultural associations, professional networks, and friendship societies that carry innocuous names — the Council for the Promotion of Peaceful National Reunification, the Chinese People’s Association for Friendship with Foreign Countries, the All-China Federation of Returned Overseas Chinese — and blur the line between state activity and private initiative. When a Chinese community organization in Manhattan turns out to be housing an undeclared police station operating on behalf of China’s Ministry of Public Security, the plausible deniability is structural. The organization looks like a community center. It functions as a transnational law enforcement outpost. Both of those things are simultaneously true.
The machinery
In 2018, Xi reorganized the UFWD to absorb the State Administration for Religious Affairs and the Overseas Chinese Affairs Office, making it the Party’s central agency for managing ethnic, religious, and diaspora affairs under one roof. The department’s director, Shi Taifeng, is a Politburo member — a level of seniority that signals the UFWD’s priority within the Party hierarchy. The UFWD oversees or coordinates with the Chinese People’s Political Consultative Conference, whose 2,000-plus members serve as the department’s primary interface with non-Party elites both domestically and abroad.
The overseas infrastructure operates through several documented channels. Confucius Institutes — Chinese-language and cultural programs embedded in universities worldwide — were launched in 2004 by Liu Yandong, who was head of the UFWD at the time. The program was funded through the CCP Propaganda Department, formally affiliated with the UFWD, and overseen by personnel based in Chinese embassies and consulates. At peak, more than 500 Confucius Institutes operated in universities globally, with over 1,000 Confucius Classrooms in secondary schools. A former Politburo Standing Committee member responsible for propaganda wrote in 2010 that China should “actively carry out international propaganda battles” on core issues and “do well in establishing cultural centers and Confucius Institutes.” Over 100 have been shut down in the United States since 2019 after the FBI flagged them as potential propaganda and intelligence-gathering platforms. Some have reopened under new names.
Chinese Students and Scholars Associations receive funding and direction from Chinese embassies and serve as monitoring and mobilization networks on foreign campuses. Multiple CSSAs have been documented suppressing academic freedom — organizing protests against speakers critical of Beijing, reporting on Chinese students’ “dissident activity” to embassy officials, and mobilizing nationalist demonstrations. The UFWD also works through overseas Chinese-language media, several of which are owned or controlled through China News Service, a UFWD-affiliated outlet.
Elite capture
The term “elite capture” describes the UFWD’s strategy for cultivating relationships with foreign decision-makers — corporate executives, university presidents, politicians, former intelligence officials — by offering access, business opportunities, paid trips to China, honorary positions, and investment partnerships. The strategy is patient and incremental. A university president accepts funding for a research center. A retired politician joins the board of a Chinese-linked foundation. A business executive receives preferential market access. None of these interactions are illegal in isolation. The aggregate effect is a network of relationships that constrains criticism of Beijing at the institutional level without any single participant necessarily understanding the full architecture they’re embedded in.
Australia became the most publicly documented case study. In the mid-2010s, investigations revealed that businessmen with close ties to UFWD-linked organizations had made significant political donations to both major Australian parties, prompting a national reckoning and new foreign interference legislation — the first of its kind in a Western democracy. The Australian Strategic Policy Institute documented the UFWD’s operational structure in a landmark 2020 report, concluding that “there’s no clear distinction between domestic and overseas united front work: all bureaus of the UFWD and all areas of united front work involve overseas activities.” New Zealand faced similar revelations. Canada launched its own public inquiry. In October 2024, Swedish journalists identified 233 individuals across Europe connected to the United Front system, and the Jamestown Foundation identified 103 UFWD-linked groups in Sweden alone, spanning culture, business, politics, and media.
The Taiwan dimension is the UFWD’s longest-running and most intensive operation. The department sponsors paid trips and summer camps to mainland China for Taiwanese youth to promote pro-unification sentiment. In August 2025, Taiwan’s Ministry of Education banned academic cooperation with three mainland universities — Jinan University, Huaqiao University, and Beijing Union University — specifically because of their documented UFWD affiliations and their role in recruiting Taiwanese students for influence activities.
What it isn’t
It’s important to distinguish the UFWD from a conspiracy theory about Chinese people. The UFWD targets the Chinese diaspora as much as it targets foreign institutions — monitoring overseas Chinese communities, suppressing dissent among Chinese nationals abroad, and co-opting community leaders to serve as intermediaries. The victims of UFWD operations include Chinese students who self-censor on foreign campuses because they know their classmates report to the embassy, Uyghur and Tibetan activists who face harassment from UFWD-linked civic organizations in their adopted countries, and Hong Kong democracy advocates who discover that community associations in their new cities are operated by the same apparatus they fled. The UFWD is a Party instrument aimed at everyone the Party considers potentially useful or potentially threatening, regardless of nationality — but with particular intensity directed at the Chinese diaspora itself.
The Council on Foreign Relations described the UFWD as an “external intelligence organization” whose officials often operate under diplomatic cover. The British government’s assessment, published in 2024, concluded that the UFWD is not an intelligence organization in the traditional sense but provides cover for Ministry of State Security officers and serves as a node in a “whole of state” approach to information gathering where the boundaries between influence, intelligence, and legitimate diplomacy are deliberately erased. That deliberate erasure — the impossibility of determining where diplomacy ends and espionage begins — is the United Front’s structural advantage and the reason it has operated for 84 years without most of the world knowing its name.
We cover the UFWD alongside BCCI’s regulatory arbitrage, the shell company architectures behind sanctions evasion, Crypto AG’s signals intelligence operation, and 20 other case studies of invisible institutional power across our Shadowcraft course — where the question isn’t whether covert influence networks exist, but how they’re built, how they’re funded, who they report to, and what the paperwork looks like when they’re finally exposed.
-
BCCI: The Most Corrupt Bank in History and How It Served Every Side of Every Conflict
The Bank of Credit and Commerce International operated in 78 countries, managed assets exceeding $20 billion, employed more than 14,000 people, and served as the personal financial institution of the CIA, Saddam Hussein, Manuel Noriega, the Medellín cartel, Abu Nidal, Pakistan’s nuclear weapons procurement network, Ferdinand Marcos, and the mujahideen fighting the Soviets in Afghanistan — simultaneously, through the same branches, often through the same officers. When regulators in seven countries raided its offices on July 5, 1991, in what remains the largest coordinated banking shutdown in history, investigators found not a bank that had been corrupted but a bank that had been designed, from its founding in 1972, as a machine for evading the laws of every country it operated in. The Kerry-Brown report to the U.S. Senate Foreign Relations Committee called it “international financial crime on a massive and global scale.” Time magazine nicknamed it the “Bank of Crooks and Criminals International.” The acting U.S. Comptroller of the Currency compared it to FTX in 2023, which is the kind of comparison that should make you realize how little has changed.
The architecture of invisibility
BCCI was founded in 1972 by Agha Hasan Abedi, a Pakistani financier who had previously built United Bank Limited before Pakistan’s nationalization wave took it from him. His new bank was incorporated in Luxembourg, headquartered in London, and majority-funded by Sheikh Zayed bin Sultan Al Nahyan, the ruler of Abu Dhabi, with Bank of America providing 25 percent of the initial capital and critical institutional credibility. From its first year, the bank was structured to be unregulable. It split itself into BCCI Holdings (Luxembourg), BCCI SA (Luxembourg), and BCCI Overseas (Grand Cayman), with parallel banks acquired or created in Geneva, Kuwait, and the Cayman Islands, layered through a web of holding companies, affiliates, subsidiaries, and nominee relationships so complex that no single regulator in any single country could see the full picture. That was the point. As the Kerry-Brown report documented, BCCI was “from its earliest days made up of multiplying layers of entities, related to one another through an impenetrable series of holding companies, affiliates, subsidiaries, banks-within-banks, insider dealings and nominee relationships.”
The growth was astonishing and unsustainable. From 19 branches in five countries in 1973 to 108 branches by 1976 to over 400 branches in 78 countries by the mid-1980s. Assets grew from $200 million to $1.6 billion in four years. Abedi pursued deposits over profits, acquiring high-net-worth clients — and high-net-worth criminals — by offering services no legitimate bank would touch. The strategy worked until it didn’t. By the late 1970s, BCCI was already secretly covering non-performing loans by creating fictional transactions and using customer deposits to fill the holes. The Abbas Gokal shipping group, BCCI’s largest borrower, was effectively bankrupt by the late 1970s. BCCI threw money at the problem and falsified the books. This carried on for 15 years.
The client list
The list of BCCI’s known clients reads like a casting call for a Cold War thriller written by someone who decided subtlety was overrated. Noriega laundered approximately $23 million through BCCI’s London branches — the bank hand-delivered him a $25,000 Persian carpet as a hospitality gesture, because when your client is a dictator who runs a country-sized drug operation, customer service matters. Pablo Escobar and other members of the Medellín cartel used BCCI for laundering. Abu Nidal, the Palestinian terrorist, used it for arms procurement. Saddam Hussein used it for weapons purchases, including a planned $110 million acquisition of 22 Argentine Mirage fighter jets arranged through BCCI’s Latin American office. Ferdinand Marcos stashed money. Hussain Muhammad Ershad, the Bangladeshi military dictator, stashed money. Samuel Doe of Liberia stashed money. If you ran a country and needed to hide the proceeds, BCCI was the institution that said yes.
But the client that makes BCCI historically significant rather than merely criminal was the Central Intelligence Agency. The CIA maintained accounts at BCCI branch offices, used the bank as a conduit for covert funding, and — according to the Kerry-Brown report and subsequent investigations — channeled billions through BCCI to the Afghan mujahideen. By 1987, CIA funding for the Afghan rebels reached $630 million annually, with Saudi Arabia matching the contribution, and much of it flowed through BCCI. The National Security Council also held accounts at the bank, used for transfers connected to Iran-Contra. A 1986 CIA memo stamped SECRET summarized the agency’s knowledge of BCCI’s activities, including the bank’s secret acquisition of First American Bankshares in Washington — a direct violation of U.S. banking law. A more detailed 30-page CIA report followed in 1989. The agency knew. The agency’s Directorate of Operations had informants inside the bank. The CIA “aggressively” targeted BCCI as an intelligence goldmine, according to deputy director Richard Kerr. And for years, nobody acted on what they found, because BCCI was too useful to shut down.
The nuclear dimension makes it worse. BCCI’s Canadian operations financed Pakistan’s procurement of nuclear weapons materials — documented in the Parvez case, where a Pakistani national attempted to acquire nuclear-related materials through the United States with BCCI financing. The CIA acknowledged in a 1991 letter to the Senate that it had reporting as early as 1987 on “BCCI being used by third world regimes to acquire weapons and transfer technology.” Libya used BCCI-connected channels for chemical weapons plant procurement. The bank wasn’t just laundering drug money. It was facilitating weapons of mass destruction procurement while the intelligence agencies that knew about it weighed the cost of shutting down an asset they were also using.
Why nobody stopped it
The regulatory failure was systemic, not accidental. BCCI had been structured from inception to split its operations across jurisdictions so that no single regulator could see the whole picture. Luxembourg saw one set of books. The Cayman Islands saw another. London saw a third. The Bank of England formed a supervisory group in 1987, but it moved slowly. U.S. regulators were warned repeatedly — by journalists, by Senate investigators, by their own agencies — and failed to act for years. Robert Mazur, a federal agent who went undercover as a wealthy businessman in Operation C-Chase, infiltrated BCCI’s private client division and documented the money laundering in real time. His operation led to the 1988 indictments that were the first serious legal action against the bank — and even that was delayed at the Justice Department’s request to avoid interfering with the sting.
The political protection was equally systemic. BCCI hired Clark Clifford — former Secretary of Defense, trusted advisor to four presidents, arguably the most connected man in Washington — to run First American Bankshares, the U.S. bank BCCI secretly and illegally controlled. Clifford and his partner Robert Altman insisted they didn’t know BCCI was behind their bank. BCCI employed lobbyists, PR firms (Hill and Knowlton), and white-shoe law firms to suppress critical coverage. One investigative journalist in the U.K., Anthony Mascarenhas, was beaten, stabbed, and had his research stolen. Abedi personally cultivated relationships with heads of state — his philosophy, as described by BCCI officer Abdur Sakhia, was to appeal to every sector: charity for Jimmy Carter, a job for Zia’s brother-in-law, deposits for central bank officials in exchange for government deposits. Suitcases of cash where necessary.
When Price Waterhouse finally audited BCCI properly in 1990, they found $1.48 billion in loans BCCI had made to its own shareholders, using BCCI stock as collateral — a circular fraud where the bank was essentially lending money to people to buy ownership of the bank that was lending them the money. The March 1991 Bank of England-ordered investigation concluded that there was “evidence of massive and widespread fraud.” The bank was shut down in July 1991 with liabilities of $10 to $14 billion. Over 6,500 depositors lost their money. Abedi, who had suffered a heart attack and retired, was never extradited. Key insiders were held incommunicado in Abu Dhabi. William Casey, the CIA director who oversaw the agency’s deepest involvement with BCCI, was conveniently dead.
What it built
BCCI didn’t invent shell company structures or nominee ownership or multi-jurisdictional regulatory arbitrage. But it proved — at a scale nobody had previously attempted — that a bank designed from inception to evade oversight could operate for nearly two decades, serve the intelligence agencies of multiple countries, finance nuclear proliferation and terrorism, launder billions in drug money, and buy political protection in the world’s most powerful capital, all without any single institution having the authority, the information, or the incentive to stop it. The tools BCCI pioneered — layered corporate structures across permissive jurisdictions, beneficial ownership concealment, regulatory fragmentation as a feature rather than a bug — are the same tools that populate the Panama Papers, the same tools that Russia’s shadow fleet uses to evade oil sanctions, the same tools that North Korea’s Lazarus Group uses to launder stolen cryptocurrency through chains of shell entities. The 2023 Corporate Transparency Act, which for the first time requires disclosure of beneficial ownership of U.S. companies, is — three decades later — a direct legislative descendant of the BCCI scandal. The question isn’t whether the reforms went far enough. It’s why it took 32 years.
We cover BCCI alongside Marc Rich’s commodity empire, the Vatican Bank, Crypto AG, Wagner Group, and 19 other case studies of covert institutional power across our Shadowcraft course — where every lecture follows the money, maps the personnel pipeline, identifies the deniability layer, and finds the moment the machinery became exposed.
-
Dolphin Signature Whistles: The Evidence That Bottlenose Dolphins Have Names
Within the first few months of life, every bottlenose dolphin develops a unique acoustic signal — a specific pattern of frequency modulations that no other dolphin in its community produces. This isn’t a generic call. It isn’t a species-wide sound. It’s an individually distinctive whistle that the dolphin will use, with minor variations, for the rest of its life. Other dolphins learn it, remember it, and — critically — copy it to get that specific individual’s attention. When researchers at the University of St Andrews played recordings of a dolphin’s own signature whistle through an underwater speaker, the dolphin called back. When they played the signature whistle of an unfamiliar dolphin, it didn’t respond. When they played the whistle of a known associate, it didn’t respond. The animal reacted specifically and exclusively to hearing its own “name” — as if someone had called it across a room.
That 2013 study, published in PNAS by Stephanie King and Vincent Janik, was the first experimental demonstration that a nonhuman mammal uses learned vocal labels to address specific individuals. The implications were immediate and significant: dolphins don’t just have identity signals the way a dog has a distinctive bark. They have signals that function referentially — labels that other dolphins can produce to mean “you, specifically.” That’s not a contact call. That’s a name.
How the system works
Signature whistles were first described by Melba and David Caldwell in the 1960s. It took decades of fieldwork — particularly from the Sarasota Dolphin Research Program in Florida, which has tracked individual dolphins since 1970 — to establish how the system operates. An infant dolphin develops its signature whistle during the first few months of life through vocal learning. The calf doesn’t inherit a whistle genetically. It listens to the whistles in its environment and constructs its own, typically by copying a whistle it heard rarely and then modifying it into something unique. The result is an individually distinctive signal that encodes identity independently of voice features — the acoustic equivalent of a name written on a nametag rather than recognized by the sound of someone’s voice.
This independence from voice cues is the detail that makes the naming analogy hold. Janik, Sayigh, and Wells demonstrated in a 2006 PNAS study that dolphins extract identity information from signature whistles even when all voice features have been removed from the recording. They synthesized whistles using computer-generated tones that preserved only the frequency contour — the shape of the whistle — and stripped everything that would tell the listener who was producing it. The dolphins still recognized the whistles. They responded preferentially to the synthetic versions of whistles belonging to individuals they knew. The contour alone carries the identity. That’s not how most animals recognize each other. Most species rely on voice cues — the timbre, the resonance, the characteristics of the individual’s vocal apparatus. Dolphins evolved a system where the pattern is the identity, not the voice. That’s structurally closer to how human names work than anything else documented in animal communication.
Copying as addressing
Dolphins don’t just produce their own signature whistles. They copy each other’s. King and colleagues showed in 2013 that copying occurs almost exclusively between animals with close social bonds — mothers and calves, allied males — and typically happens when the animals are separated and apparently trying to reunite. One pair of allied males was recorded copying each other’s whistles 12 years apart, preserving the fine acoustic details across more than a decade. Signature whistles make up roughly 50 percent of all whistles a dolphin produces, making them by far the most common sound in the repertoire.
The copying is selective and precise but not exact. When a dolphin copies another’s whistle, it introduces minor but consistent modifications — subtle enough to preserve the referential content (whose whistle this is) while potentially marking it as a copy rather than the original. This is a nuance researchers are still working to understand. It’s possible the modifications function like quotation marks — a way of saying “I’m producing your name” rather than “I am you.” If that interpretation holds, it would mean dolphins are not just labeling individuals but doing so with a meta-communicative marker that distinguishes original production from quotation. That’s a level of communicative sophistication that, as of 2026, hasn’t been fully confirmed but also hasn’t been ruled out.
Male bottlenose dolphins in Shark Bay, Australia, retain individual vocal labels even within multi-level alliance structures — coalitions of two to three males that cooperate to herd females, embedded within larger super-alliances of up to 14 males. King and colleagues published in Current Biology in 2018 that allied males maintain their individually distinctive signature whistles rather than converging on a shared group call, which is what you’d expect if the whistles served a group-identity function. The fact that they don’t converge — that each male keeps his own whistle even within a tightly bonded coalition — supports the interpretation that the whistles are individual labels, not team jerseys.
Motherese
In 2023, a study published in PNAS by Sayigh and colleagues from the Sarasota Dolphin Research Program demonstrated something that stopped a lot of people scrolling: dolphin mothers modify their signature whistles when their calves are present. The modifications — shifts to higher maximum frequencies — parallel the acoustic changes human parents make when speaking to infants, the phenomenon known as “motherese” or infant-directed speech. Human motherese involves higher pitch, wider pitch range, and exaggerated intonation. Dolphin motherese involves higher-frequency whistles with extended contours. Same function, different species, different medium.
The finding matters because it suggests that the modification isn’t a side effect of arousal or environment — mothers don’t shift their whistles when other dolphins are present, only when their own calves are nearby. The adjustment is calf-directed. Whether it serves the same developmental function as human motherese — facilitating attention, bonding, and potentially vocal learning — remains an open question. But the structural parallel is hard to dismiss.
Beyond signature whistles
The most recent advance — a 2025 preprint from Sayigh, Janik, and the Sarasota team — moves past signature whistles entirely into territory that may prove even more significant. Having catalogued the signature whistles of most individuals in a community of 170 dolphins, the researchers are now documenting “non-signature whistles” — stereotyped whistle types that are not individually distinctive but are shared across multiple animals. They’ve identified 22 shared non-signature whistle types so far, two of which have been produced by at least 25 and 35 different dolphins respectively. If signature whistles are names, non-signature whistles may be something closer to words — shared acoustic signals with community-wide meaning rather than individual identity. Playback experiments filmed with drones are underway to determine what these shared whistles mean and how dolphins respond to them. The work was selected as a finalist for the Coller-Dolittle competition, which features non-invasive approaches to studying animal communication.
Deep-learning classifiers are also being developed to automate signature whistle identification — a task that previously required expert human listeners to visually compare spectrograms. Jensen and colleagues published methods in 2024 for training neural networks to classify signature whistles from field recordings, which could turn the Sarasota whistle database into a passive population-monitoring tool. Hydrophone networks throughout Sarasota Bay could, in principle, track individual dolphins by their whistles the way cell towers track phones by their signals.
The comparative picture
Dolphins are no longer alone in the naming evidence. In 2024, a study published in Nature Ecology & Evolution demonstrated that African elephants address one another with individually specific name-like calls — not by copying, as dolphins do, but by producing arbitrary learned labels, which is structurally even closer to how human names work. A separate 2024 study in Science showed vocal labeling in marmoset primates. The evidence for animal naming has gone from a single-species curiosity to a cross-taxon pattern in two years.
But dolphins remain the most extensively documented case, with 50 years of signature whistle research, a longitudinal dataset spanning decades of known individuals, and a level of experimental rigor — playback studies with synthetic whistles, controlled for voice cues, replicated across wild and captive populations — that the elephant and marmoset findings don’t yet match. The combination of vocal learning — the rare ability to hear a sound and reproduce it, shared by dolphins, parrots, songbirds, hummingbirds, bats, and humans but absent in most mammals — with the social complexity of fission-fusion groups, where individuals constantly separate and reunite, created the evolutionary pressure for a labeling system. When you can’t see your allies in murky water, you need a way to call them by something more specific than “hey.”
The question the field is converging on isn’t whether dolphins have names. The evidence for that is now robust. The question is how much further the communication system extends beyond naming — whether the shared non-signature whistles represent a rudimentary vocabulary, whether the modifications during copying carry grammatical information, and whether the dolphin communication system has more structure than we’ve been able to decode. The Neurozoology course covers dolphin signature whistles alongside octopus distributed cognition, corvid tool use and funerary behavior, and electroreception in sharks and platypuses — the full catalog of neural capabilities that evolution produced outside the human lineage, most of which we didn’t know existed until someone thought to look.
-
Room-Temperature Superconductors: What Happened After LK-99 and Where the Search Stands
In July 2023, a small Korean lab called Q-Centre posted two preprints claiming a material called LK-99 — a copper-doped lead apatite — was a room-temperature, ambient-pressure superconductor. The internet lost its mind. Stock markets moved. Twitch streamers watched replication attempts live. A Chinese researcher uploaded a levitation video to Bilibili that got 4.5 million views in nine hours. For approximately two weeks, it felt like humanity might have stumbled into the most transformative materials discovery of the century — a substance that conducts electricity with zero resistance at room temperature and normal pressure, which would transform power grids, computing, transportation, medical imaging, and essentially every system that moves electrons through wire.
It wasn’t real. Within a month, labs worldwide had synthesized LK-99 and found no superconductivity. No zero resistance. No Meissner effect. No flux pinning. The partial levitation in the original Korean video turned out to be a ferromagnetic impurity — copper sulfide — not a superconducting phenomenon. A comprehensive rebuttal by Georgescu and colleagues, updated in early 2025 and published in Chemistry of Materials, dismantled the original claims point by point. LK-99 was a semiconductor with interesting magnetic properties. It was not a superconductor at any temperature.
But LK-99 was only the most public failure. The field’s deeper wound came from Ranga Dias.
The Dias fraud
Ranga Dias, a physicist at the University of Rochester, published a paper in Nature in October 2020 claiming room-temperature superconductivity in a carbonaceous sulfur hydride under extreme pressure — roughly 267 gigapascals, the kind of pressure found near Earth’s core. Nature had actually ignored the majority opinion of its peer reviewers, who expressed serious concerns. The paper launched Dias to fame. Rochester doubled his salary. Venture capitalists courted his startup, Unearthly Materials, which raised $17 million to commercialize the discovery. Nobody could reproduce the results.
In 2022, physicist James Hamlin at the University of Florida discovered that a section of the magnetic susceptibility data in Dias’s paper appeared to have been copied and pasted from one temperature range to another. Hamlin also found that Dias had plagiarized portions of his PhD thesis from Hamlin’s own earlier thesis. Nature retracted the 2020 paper in September 2022.
Dias responded by publishing another room-temperature superconductor claim in Nature in March 2023 — nitrogen-doped lutetium hydride, this time at near-ambient pressure. The paper went through extra review. It didn’t matter. Other labs couldn’t replicate the results. Dias’s own graduate students contacted Nature with concerns about data validity. That paper was retracted in November 2023. A third retraction followed in Physical Review Letters. Then a fourth, then a fifth.
In March 2024, a University of Rochester investigation — conducted by external physicists at the National Science Foundation’s request — concluded that Dias had engaged in “falsification, fabrication, and/or plagiarism of data, images, and text.” He was stripped of his students and laboratories. As of November 2024, he is no longer employed at Rochester. Five papers retracted. Millions in grants. A $17 million startup. And not a single reproducible result. As Lilia Boeri at Sapienza University of Rome put it, Dias’s “inconsiderate behaviour has harmed the reputation of the field.” James Hamlin, the physicist who caught the fraud, was more direct: the saga is “damaging to science in general, and superconductivity research more so.”
Where the field actually stands
Strip away LK-99 and Dias, and the legitimate science of superconductivity is in a more interesting place than the fraud cycle suggests. The highest confirmed superconducting temperature at ambient pressure remains around 135 Kelvin (-138°C), achieved in cuprate superconductors — the copper-containing ceramics discovered in 1986 that earned Bednorz and Müller the Nobel Prize. Under extreme pressure, lanthanum decahydride superconducts at 250 Kelvin (-23°C) at 150 gigapascals — genuinely close to room temperature, but at pressures that require diamond anvil cells and have no practical application.
The real action is in nickelates. In February 2025, researchers at SLAC National Accelerator Laboratory and Stanford achieved a breakthrough that got far less public attention than LK-99 but matters considerably more: they stabilized a nickelate superconductor at room pressure for the first time. Nickelates are chemically similar to cuprates — the same class that holds the ambient-pressure temperature record — and had previously shown superconducting behavior only under extreme pressure. The SLAC team demonstrated that lateral compression from a substrate could stabilize the material without the diamond anvil cells that make high-pressure experiments impractical. This doesn’t mean nickelates superconduct at room temperature. They don’t, not yet. But it means researchers can now study them using advanced techniques like X-ray scattering that were impossible when the materials only existed under crushing pressure. The constraint has shifted from “can we make it at all” to “can we understand it well enough to improve it.”
A separate advance came from Penn State in October 2025, where researchers used a framework called zentropy theory — merging statistical mechanics with quantum physics and computational modeling — to predict superconducting behavior from a material’s electronic structure. The approach correctly identified known superconductors and offered a method for screening candidate materials computationally rather than synthesizing thousands of compounds by trial and error.
Then in March 2026, a programmatic research agenda published in the Proceedings of the National Academy of Sciences by a multi-institutional team laid out the case for a coordinated global push toward room-temperature superconductivity. The paper noted that no fundamental physical laws prevent it — the barrier is engineering and materials science, not physics. Recent advances in pressure quenching of the cuprate Hg-1223 achieved a record critical temperature of 151 Kelvin at ambient pressure. The authors argued that combining ab-initio computational simulations — now capable of modeling materials at the nanometer scale, a tenfold improvement over capabilities just a few years ago — with machine learning and AI-driven materials screening could systematically push critical temperatures higher. The paper reads less like a research summary and more like a call to arms: join forces worldwide, use modern computational tools, and treat room-temperature superconductivity as an engineering program rather than a lottery ticket.
Why it matters and why it’s hard
A room-temperature, ambient-pressure superconductor would eliminate the roughly 5 percent of electricity lost in transmission across the U.S. grid alone — a figure worth tens of billions of dollars annually and a meaningful reduction in carbon emissions without building a single new power plant. MRI machines, which currently require expensive liquid helium cooling for their superconducting magnets, could operate without cryogenics — relevant to anyone who’s read about the helium shortage and the supply chain fragility that makes every MRI refill a logistics problem. Maglev trains, which already use superconducting technology in Japan’s SCMaglev system, could become economically viable for mass transit instead of remaining engineering showcases. Particle accelerators, fusion reactors — where superconducting magnets are the enabling technology for plasma confinement — and quantum computers that currently require millikelvin temperatures to maintain superconducting qubits could all be transformed.
The difficulty is that we don’t fully understand how high-temperature superconductivity works. In conventional superconductors, the mechanism — electrons forming Cooper pairs through interaction with the crystal lattice — is well understood and was described by Bardeen, Cooper, and Schrieffer in 1957. But the cuprate superconductors that hold the temperature record don’t follow this mechanism. Something else is creating the electron pairing, and after nearly four decades of research, there’s no consensus on what it is. You can’t engineer your way to a higher critical temperature when you don’t have a complete theory for why the current record holders work. The nickelate breakthrough matters precisely because it gives researchers a new family of materials to study alongside cuprates — same neighborhood of the periodic table, different structure, potentially different mechanism — which means more data points for the theorists to work with.
The fraud problem is also a structural problem
LK-99 and Dias didn’t happen in a vacuum. The incentive structure of academic science — where a single Nature paper can double a salary, launch a startup, and generate millions in grant funding — creates enormous pressure to produce extraordinary results. Room-temperature superconductivity is the most sought-after prize in condensed matter physics. The gap between “promising measurement” and “confirmed superconductor” requires demonstrating zero resistance, the Meissner effect, flux pinning, a temperature-dependent critical field and current, and a specific heat anomaly. LK-99’s original papers demonstrated none of these. Nature published Dias’s first claim over reviewer objections. The field’s quality-control mechanisms failed at every level, from peer review to institutional investigation — Rochester cleared Dias three times before external reviewers found the fraud.
The March 2026 PNAS roadmap addresses this implicitly by calling for tighter integration between theory, computation, and experiment — essentially arguing that the field needs to stop waiting for someone to stumble onto a miracle material and start engineering candidates systematically. The AI-driven materials screening approach treats the problem the same way pharmaceutical companies treat drug discovery: model the candidates computationally, screen for promising properties, synthesize the top candidates, test rigorously, iterate. It’s less romantic than a eureka moment. It’s also less susceptible to fraud, because the computational predictions are independently verifiable before anyone walks into a lab.
Room-temperature superconductivity occupies a strange position: it’s the moonshot where the physics explicitly permits success, the engineering hasn’t delivered it, and the most famous recent claims turned out to be fabricated. The legitimate science — nickelate stabilization at SLAC, zentropy theory at Penn State, AI-accelerated materials screening, the 151 K ambient-pressure record in Hg-1223 — is advancing on a timeline measured in decades, not press cycles. We cover it alongside solid-state batteries, fusion energy, and space-based solar power in our Moonshot 2169 course — where each technology gets the same treatment: define what “done” actually means, name the constraints that prevent it, separate the press releases from the physics, and be honest about how far the gap between laboratory demonstration and deployable technology really is.
