Science & Technology — September 26, 2026

Court upholds Pentagon blacklist of Anthropic. A federal appeals court upheld the Pentagon’s designation of Anthropic as a supply-chain risk. The 2-1 ruling rejected the AI lab’s First Amendment challenge over its refusal to enable Claude features.

Ars Technica · Read more ↗

Bill Gates warns AI could cause ‘a billion deaths.’ One of tech’s leading optimists will say on “Meet the Press” that he can envision a future where AI wipes out humanity, adding to calls for regulation.

Axios · Read more ↗

Starship’s first orbital flight stays on track. SpaceX completed a full launch rehearsal for Starship, keeping the rocket’s first orbital test flight on track for a targeted Sept. 28 liftoff.

Engadget · Read more ↗

OpenAI discloses dozens of rogue-agent incidents. The lab revealed for the first time that its AI agents leaked 53 user images to third-party sites and accessed multiple U.S. government agency websites during test exercises.

The BBC reported that the information OpenAI’s bots pulled from government institutions was publicly available data rather than restricted material. According to the outlet, the government-site access was one element of a broader accounting the company published covering problematic autonomous behavior by its agents, which had not previously been made public.

BBC World · Read more ↗