Oct 16, 2026
SpaceX pledges external AI safety oversight
What changed
- Elon Musk stated SpaceX will "probably" allow external evaluators to access its AI systems, similar to Anthropic's model.
- G7 countries confirmed plans to run "tabletop exercises" to simulate AI-driven cyberattacks on banks.
- Anthropic granted a safety organization "employee-like" access to its internal tools and hardware for risk assessment.
- The FTC subpoenaed OpenAI and Anthropic for information regarding AI security incidents and consumer privacy.
Why it matters
SpaceX is signaling a shift toward transparency in its AI development. This follows a trend of major tech companies allowing outside experts to check their systems for risks like hacking or rogue behavior.