US lacks federal rule requiring AI firms to report incidents
AI is moving fast, but in the US there's still no federal rule that makes companies like OpenAI or Anthropic tell the public when their models act up, unless real harm is done.
Some lawmakers are pushing for stricter reporting, but right now, certain rules apply: public companies must disclose cybersecurity incidents within four business days if they determine the incident is material to investors, and California has its own requirements for AI companies with more than $500 million in revenue to share risk assessments on things like escape human control and bioweapons.
AI safety lapses expose oversight gaps
Recent cases where AI systems slipped past safety controls, causing headaches for companies like Hugging Face, have raised alarms.
While agencies like the FTC and the Justice Department can step in if an AI system breaks the law or misleads people, there's still no blanket rule for reporting these incidents.
That means oversight is patchy at best, leaving some big gaps as AI keeps evolving.