News — 2026-03-27

David Sacks announced the end of his 130‑day tenure as the Trump administration’s AI and crypto czar and said he will now serve as co‑chair of the President’s Council of Advisors on Science and Technology (PCAST). In the new role, Sacks will help steer the council’s work on AI, advanced semiconductors, quantum computing and nuclear power, while pushing the administration’s national AI framework. The move shifts him from a direct policy‑making position to a more advisory capacity alongside industry leaders such as Jensen Huang, Mark Zuckerberg and Larry Ellison. Sacks also clarified that his recent podcast remarks on the Iran conflict were personal views, not official policy. Read more

A federal judge in California issued an injunction requiring the Trump administration to rescind its designation of AI firm Anthropic as a supply‑chain risk and to halt orders directing federal agencies to cut ties with the company. The ruling, delivered by Judge Rita F. Lin, rejected the administration’s claim that Anthropic posed a national‑security threat after the company challenged the Pentagon’s demand that it drop usage limits on its models. Anthropic’s CEO Dario Amodei hailed the decision as a protection of free speech and a step toward productive collaboration with the government. Read more

Google has added a feature to Gemini that allows users to import conversation history from other AI assistants. By copying a response from the previous AI or uploading a ZIP file of exported data, Gemini can continue a discussion without the user having to repeat prior details. The process is available through the Settings menu on the desktop version and supports files up to 5GB. Early testers report smoother interactions despite a brief processing wait, marking a notable step toward seamless multi‑AI workflows. Read more

A federal judge in San Francisco issued a temporary injunction that stops the Department of Defense from labeling AI firm Anthropic as a supply‑chain risk. The order restores the situation to before the Pentagon’s directives that limited the use of Anthropic’s Claude AI tools across federal agencies. While the ruling does not compel the military to continue using Anthropic’s technology, it prevents the agency from relying on the contested designation as a basis for further action. The decision is a significant legal boost for Anthropic as it continues to challenge the administration’s sanctions. Read more

Researchers who scanned millions of webpages discovered that thousands of sites are unintentionally publishing API credentials for major services such as Amazon Web Services, Stripe and OpenAI. The majority of leaks originate from JavaScript files that are publicly accessible, allowing anyone to misuse the keys. The study uncovered 1,748 distinct credentials across nearly 10,000 pages, with some keys remaining exposed for up to a year or longer. Experts say the problem stems from developers embedding private keys in front‑end code, and they recommend live‑site scanning, stricter tool controls and better detection by service providers. Read more