Anthropic said that "no testers have yet been able to find a universal jailbreak — a jailbreak method that can very broadly bypass the model's safeguards, unblocking a wide range of cyber capabilities," and added that total avoidance of any jailbreaks is not currently possible for Anthropic or any other company. This is relevant because the June 2026 export-control controversy centered on concerns about bypassing safeguards in Anthropic models, not on Kimi K3 distilling a model that had not yet been completed.

Kimi K3 was described as a powerful model released by Moonshot, and the reporting noted that U.S. officials and AI executives were warning about China's propensity to rip off technology through unauthorized "distillation". The article also said Anthropic had previously accused Chinese labs of using distillation to steal its AI technology, but it does not describe a congressional question about Kimi K3 distilling a not-yet-completed Anthropic model.

On February 23, Anthropic said it identified industrial-scale distillation campaigns by three China-based AI labs — DeepSeek, Moonshot AI (known for its Kimi models) and MiniMax — that collectively generated more than 16 million exchanges with Claude through approximately 24,000 fraudulent accounts, in violation of the company’s terms of service and regional access restrictions. The campaigns, Anthropic said in its blog post, used a consistent technical architecture: fraudulent account clusters routed through commercial proxy services that resell API access to Claude and other frontier models at scale. Anthropic said it is responding with behavioral fingerprinting classifiers designed to detect coordinated distillation patterns, including chain-of-thought elicitation.

The Committee on Homeland Security (Committee) is closely examining how advances in quantum computing, artificial intelligence (AI), and cloud computing impact cybersecurity and national security. In light of these developments, and recognizing the central role Anthropic plays in detecting and disrupting this activity, we request that you testify at a joint hearing before the Committee’s Subcommittee on Cybersecurity and Infrastructure Protection and Subcommittee on Oversight, Investigations, and Accountability titled, “The Quantum, AI, and Cloud Landscape: Examining Opportunities, Vulnerabilities, and the Future of Cybersecurity,” on Wednesday, December 17, 2025, at 10:00 a.m. in 310 Cannon House Office Building. The letter is signed by Chairman Mark E. Green, Ranking Member Eric Swalwell, and Ranking Member Shri Thanedar; it does not contain any reference to Kimi, K3, or questions about distilling unfinished Anthropic models.

Kimi K3, the largest open-weight model ever released, reportedly performed "competitively" with Fable 5 and "substantially outperformed" Anthropic's Opus 4.8 and OpenAI's GPT 5.6 Sol, according to Moonshot. The reporting says many observers, including Dario Amodei, did not expect a Chinese AI lab to release a model that could approach the U.S.'s best offerings for another six months.

Anthropic, an American artificial intelligence firm, has accused Alibaba, the prominent Chinese tech and e-commerce conglomerate, of unlawfully siphoning capabilities from its Claude AI model. The incident is referred to by Anthropic as a "distillation" operation, where a less powerful model is trained using the outputs from a more advanced one. The correspondence, dated June 10, was directed to Senators Tim Scott and Elizabeth Warren, who serve as chair and member of the U.S. Senate Banking Committee, in advance of an upcoming hearing focused on AI.

This CyberScoop report describes how the House Homeland Security Committee sent a letter calling on Anthropic CEO **Dario Amodei** to testify about a likely Chinese espionage campaign that used the company’s Claude AI tool. It explains the committee’s concerns about national security and AI use in cyber operations. The article **does not reference Kimi, Kimi K3, or any question by a member of Congress about distilling an unfinished Anthropic model**, focusing instead on misuse of Claude by threat actors and potential oversight hearings.

A company account said the U.S. government ordered Anthropic's model pulled after concerns that a China-linked group had gained access, raising the prospect of Beijing reverse-engineering the model through distillation. That is the closest directly relevant reporting found here, but it concerns alleged access to an existing model, not a congressional question about a model that had not yet been completed.

Amodei told the Senate Judiciary Subcommittee that companies developing frontier AI systems should comply with stringent cybersecurity standards, and that there should be a "testing and auditing regime" for new and more powerful models. This establishes his public congressional testimony on model safety, but it does not mention Kimi K3 or any unfinished Anthropic model.

"Distillation" is a term used for a technique whereby outputs from a more advanced AI model (the "teacher") are used to improve a less advanced AI model (the "student"). This approach was originally developed as a way to extract ("distill") knowledge and capabilities from a larger model (or models) in order to create a smaller model with comparable knowledge and capabilities. In recent months, the three leading U.S. AI companies (OpenAI, Google, and Anthropic) have released information about so-called "distillation attacks" they detected on their services, carried out by Chinese AI firms. The most detailed report (from Anthropic) describes three different Chinese firms (DeepSeek, Moonshot AI, and MiniMax) using over 24,000 accounts for over 16 million interactions that Anthropic judged to be distillation attempts.

Anthropic announced it had received an export control order that mandated suspension of access to its newest models, Claude 5 and Mythos 5, for all foreign nationals, including those employed by Anthropic. The article says this directive followed Amodei's public call for stricter, enforceable AI regulation, but it does not support the claim that a member of Congress asked him how Kimi K3 could distill an unfinished Anthropic model.

On June 10, Anthropic addressed a letter to Senators Tim Scott and Elizabeth Warren alleging that the Chinese tech conglomerate Alibaba has been covertly imitating its A.I. technologies through a method known as distillation. According to the correspondence, which was reviewed by The New York Times, Alibaba, similar to other Chinese enterprises, accessed Anthropic's technologies by creating tens of thousands of unauthorized accounts. The company then utilized the gathered data to enhance its own A.I. systems. Anthropic requested that the senators, who oversee a Senate committee poised to conduct a hearing on A.I., investigate measures to limit China's distillation practices.

Four senators sent a letter to Defense Secretary Hegseth and Dario Amodei urging both sides to calm tensions and extend negotiations beyond the Pentagon's "hasty" deadline. The reporting confirms continued congressional engagement with Amodei on AI policy, but it does not contain the specific Kimi K3 distillation claim.

On Wednesday, a closed-door briefing hosted by the bipartisan House Homeland Security Committee featured Jack Clark from Anthropic, according to sources with knowledge of the event. The primary topics of discussion included model distillation—transforming extensive AI models into more compact versions—and export regulations. The conflict with the Pentagon was not a major point of emphasis, according to two sources. Background: In December, the committee conducted a hearing on the future of artificial intelligence and cybersecurity, featuring witnesses from Anthropic, Google Quantum X, and Seven Hill Ventures.

This Politico piece examines a political dispute over U.S. export controls as they apply to Anthropic and its Claude models. It reports on how Anthropic personnel, including CEO Dario Amodei, have been in discussions with U.S. officials about serving government and military contracts and about restrictions on exporting advanced AI systems. The article **does not mention Kimi, Kimi K3, or any line of questioning from a member of Congress about model distillation or distilling an unfinished Anthropic model**; its focus is export control policy and lobbying dynamics.

A 60 Minutes transcript noted that Congress has not passed legislation requiring commercial AI developers to conduct safety testing, leaving companies and their leaders largely to police themselves. This is useful context for Amodei's interactions with Congress, but it does not mention Kimi K3 or an unfinished Anthropic model.

Anthropic, the San Francisco-based creator of the Claude large-language model, sent a June 10 letter to Senate Banking Committee chair Tim Scott and ranking member Elizabeth Warren warning that operators linked to Alibaba’s Qwen AI lab had conducted what the company calls the “largest known distillation campaign” against Claude to date. The letter, seen by Reuters, details a three-week effort from April 22 to June 5, 2026 that generated more than 28.8 million exchanges with Claude through nearly 25,000 fraudulent accounts. Nevertheless, the evidence base remains limited to Anthropic’s internal logs and the letter to U.S. senators. No independent technical audit has been published, and Alibaba’s silence leaves a verification gap.

This Council on Foreign Relations event, "The Future of U.S. AI Leadership" with Anthropic CEO Dario Amodei, features a moderated discussion about AI, national security, and policy. In the transcript, Amodei talks about security, deployment, and possible safety mechanisms like allowing models to "quit" a job. The discussion is with policy experts and not a congressional hearing, and there is **no reference to Kimi, Kimi K3, or questions about distilling Anthropic models, completed or otherwise**, by U.S. lawmakers.

This CBS News piece provides the full transcript of an interview with Anthropic CEO Dario Amodei regarding the company’s work with the U.S. government, including national security applications and tensions with a renamed "Department of War" over serving models for military use. Amodei discusses ultimatums, the Defense Production Act, and red lines for deploying Claude to government clients. The transcript **does not contain any discussion of Kimi, Kimi K3, or questions from members of Congress about how another company could distill an unfinished Anthropic model.**

Anthropic quietly sat down with the House Homeland Security Committee for a closed-door session that zeroed in on some of AI's thorniest policy questions. The big topics: model distillation — the process of compressing large AI systems into smaller, more portable versions — and export controls. Both issues sit at the intersection of national security and AI competitiveness, and lawmakers apparently wanted details.

The transcript states that Amodei discussed what Congress should do as artificial intelligence accelerates faster than Washington is ready for, including transparency legislation and addressing authoritarian adversaries. This confirms a congressional-policy setting for Amodei's remarks, but it does not substantiate the specific claim about Kimi K3 distilling an Anthropic model not yet completed.

Anthropic co-founder Jack Clark held a closed-door briefing with bipartisan members of the House Homeland Security Committee on Wednesday. Talks centered on model distillation – compressing large AI systems into smaller ones – and export controls, not the ongoing Pentagon lawsuit. Anthropic is currently suing the federal government after being designated a supply chain risk, but that dispute was not a central focus. The tone was described as 'friendly' by sources familiar with the meeting; specifics remain undisclosed due to the closed format.

U.S. politicians and several major U.S. AI companies including Anthropic and OpenAI have accused Chinese AI models of illicit 'distillation' of their models to extract their technologies, a claim that Beijing says is 'groundless.' Anthropic in February accused DeepSeek, Moonshot and a third China-based AI lab, MiniMax, of engaging in campaigns to 'illicitly extract Claude's capabilities to improve their own models' using the distillation technique that 'involves training a less capable model on the outputs of a stronger one.' Anthropic said that distillation can be a legitimate way to train AI systems but it's a problem when competitors 'use it to acquire powerful capabilities from other labs in a fraction of the time, and at a fraction of the cost, that it would take to develop them independently.'

This Lex Fridman podcast episode features a long-form conversation with Dario Amodei on Claude, AGI, and AI safety. Amodei describes model autonomy levels, deployment considerations, and his philosophy on building AI systems. The episode **does not involve U.S. lawmakers and does not mention Kimi or Kimi K3, nor any congressional questions about distilling Anthropic models**, focusing instead on general AI issues and Anthropic’s approach.

On February 23, 2026, Anthropic published a detailed account titled “Detecting and Preventing Distillation Attacks,” accusing three Chinese AI laboratories — DeepSeek, Moonshot AI (maker of the Kimi models), and MiniMax — of running what it called industrial-scale extraction campaigns against Claude. The company’s own figures, corroborated independently by Bloomberg, CNBC, and the South China Morning Post, were specific: roughly 24,000 fraudulent accounts were used in total, in violation of Anthropic’s terms of service and regional access restrictions. In a June 10, 2026 letter to the U.S. Senate Banking Committee, Anthropic alleged that operators affiliated with Alibaba’s Qwen division had run the largest distillation campaign it had ever disclosed: roughly 25,000 fraudulent accounts generating 28.8 million exchanges with Claude between April 22 and June 5, 2026, targeting software engineering, agentic reasoning, and cybersecurity capabilities.

This Times of India article reports on a controversy where the U.S. government ordered Anthropic to pull an unspecified AI application, giving the company a 90‑minute window, which Anthropic called disproportionate. It notes that Anthropic described the incident as a misunderstanding and was working to restore access. The story **does not mention any congressional hearing, any member of Congress, Kimi or Kimi K3, or questions about how Kimi K3 might distill an incomplete Anthropic model**; it centers on content takedown and regulatory tensions.

Anthropic has accused Alibaba of running the largest model distillation campaign it has ever publicly disclosed — and it made the accusation not in a press release, but in a June 10, 2026 letter to the US Senate Banking Committee. According to that letter, operators affiliated with Alibaba and its Qwen AI lab generated more than 28.8 million exchanges with Claude through almost 25,000 fraudulent accounts. On June 10, 2026, Anthropic wrote to Senator Tim Scott, chair of the US Senate Banking Committee, and Senator Elizabeth Warren, its ranking member, alleging that operators affiliated with Alibaba’s Qwen AI lab had run what it called the largest known distillation attack against Anthropic to date. Per the letter, the alleged campaign ran from April 22 to June 5, 2026 — a 44-day window — and targeted Claude’s software engineering, agentic reasoning and cybersecurity capabilities, specifically the abilities embodied in Anthropic’s Mythos Preview frontier model.

This Future of Life Institute podcast episode features Daniela and Dario Amodei discussing Anthropic’s mission and AI safety. They talk about building reliable, interpretable, steerable AI systems and describe their research directions. The conversation is not a congressional setting and does **not** involve any mention of Kimi, Kimi K3, model distillation by external firms, or questioning by members of the U.S. Congress.

A widely circulated Hacker News thread discusses a "Statement from Dario Amodei on our discussions with the Department of War" where Amodei alleges threats to invoke the Defense Production Act and label Anthropic a supply chain risk. The thread centers on U.S. government pressure over military uses of Anthropic’s models. In the comments and linked statement, there is **no indication of any member of Congress asking about Kimi K3 or about distilling an Anthropic model that had not yet been completed**; the focus is executive-branch pressure, not a specific congressional exchange about Kimi or model distillation.

The available search results do not show a primary-source congressional transcript or hearing record containing the quoted question about Kimi K3 distilling an Anthropic model that had not yet been completed. The closest primary source in the set is Amodei's 2023 Senate testimony on AI safety and testing.