Anthropic

The UK is actively seeking to deepen ties with AI company Anthropic, contrasting with the US which sanctioned them for refusing to compromise on ethical AI guardrails. London aims to be a welcoming regulatory environment, offering proposals like a dual stock listing to attract Anthropic. This move highlights the UK’s strategy to position itself as a balanced leader in AI governance, valuing ethical development while fostering innovation, and competing for global AI talent.

Broadcom has secured significant AI chip manufacturing deals with Google and expanded its partnership with Anthropic. These agreements highlight the intense demand for AI infrastructure to power generative AI models. Broadcom’s role in manufacturing Google’s AI processors and supplying compute capacity to Anthropic positions it as a key player in the rapidly growing AI market.

Generative AI firm Anthropic experienced a turbulent week following a significant internal data leak, involving source code for its Claude Code assistant and details of its unreleased “Mythos” AI model. While Anthropic attributes the issues to human error and assures no sensitive customer data was compromised, the leaks expose proprietary technology and raise concerns about its security protocols amidst intense competition. These incidents highlight the challenges of safeguarding complex AI codebases and will test Anthropic’s operational resilience.

Cybersecurity stocks plummeted following reports of Anthropic’s advanced AI model, “Mythos,” which possesses enhanced cyber capabilities but also potential security vulnerabilities. The news, highlighting the dual-use nature of sophisticated AI, led to significant drops in major cybersecurity firms’ stock prices. This development underscores the evolving threat landscape and the challenges facing the sector in adapting to AI’s disruptive influence.

A federal judge has granted AI startup Anthropic a preliminary injunction against the Trump administration’s blacklisting, pausing the Pentagon’s designation of Anthropic as a supply chain risk. The ruling effectively allows Anthropic to continue its work with federal agencies while litigation proceeds. The judge criticized the administration’s actions as illegal First Amendment retaliation, stating the government’s stance was an “Orwellian notion.” Anthropic expressed gratitude and a commitment to working productively with the government.

A federal judge is considering whether the Pentagon unlawfully banned Anthropic’s AI models, calling the move an attempt to cripple the company. Anthropic argues the designation as a “supply chain risk” is retaliatory and lacks basis. The judge questioned if Anthropic faced adverse actions for critiquing government contracting and whether a low threshold was used for the designation. A ruling is expected soon, which could allow Anthropic to continue government contractor relationships while its lawsuit proceeds.

Anthropic’s Claude chatbot now autonomously performs tasks on user computers, using smartphone commands. This leap challenges emerging AI agents like OpenClaw by enabling Claude to open applications, browse the web, and manage files. While Anthropic emphasizes safeguards and user permission, this advancement highlights the industry’s push towards AI agents capable of continuous, autonomous operation. Dispatch further integrates Claude into professional workflows.

Senator Elizabeth Warren criticizes the Pentagon’s classification of AI startup Anthropic as a “supply chain risk,” calling it retaliation. She fears the DoD seeks tools for domestic surveillance and autonomous weapons without adequate safeguards. This occurs amidst heightened scrutiny of DoD tech contracts, especially concerning Anthropic’s dispute over AI usage and OpenAI’s similar agreement, which also raises concerns about potential misuse for surveillance and autonomous weapons. Warren demands transparency on these deals, highlighting a conflict between national security, technology, and oversight.

The Pentagon has labeled Anthropic’s AI models as a “supply chain risk” due to concerns over embedded policy preferences compromising defense integrity. This unprecedented move requires contractors to certify non-use of Anthropic’s Claude models. Anthropic is suing the U.S. government, citing irreparable harm and jeopardized contracts. The Department of Defense states the action isn’t punitive, acknowledging a transition period and existing usage by contractors like Palantir.

Palantir continues to integrate Anthropic’s Claude AI models into its products, despite the Pentagon designating Anthropic a supply-chain risk. CEO Alex Karp confirmed the current integration, noting future plans to support other large language models. The Department of Defense faces challenges phasing out deeply embedded systems, with potential exemptions for mission-critical activities. This situation highlights the complex interplay of national security, technological integration, and regulatory scrutiny in defense AI.