FTC Demands AI Safety Records from OpenAI and Anthropic

The Federal Trade Commission (FTC) has sent 50-page demands to OpenAI and Anthropic, asking for records on security incidents, safeguards, marketing, and privacy. This is part of a broader effort to regulate AI and ensure companies are transparent about safety. For you, this could mean safer AI products and more accountability, but also potential delays in new features as companies adjust to stricter rules.

Meta's new business terms, effective October 30, allow it to scrape and train AI on publishers' websites. This means your favorite news sites and blogs could be used to improve Meta's AI without direct payment to the publishers. While this might lead to smarter AI, it raises concerns about copyright and fair compensation for content creators.

In a surprising twist, two of the largest open-source AI models now fit on a desktop computer. Meta's Muse Spark Open and Nvidia's Nemotron 3.5 Super VL are powerful models that can run locally on hardware you own, like a high-end Mac. This could save businesses from variable cloud costs, turning them into fixed hardware expenses. However, these models may not match the performance of top-tier paid AI, so testing is needed before switching.

Other news: Character.ai filed for bankruptcy, Citizen Lab traced spyware through targeted ads, and Accenture now funds $29/month of Claude per employee. Bank of America hired 1,000 contractors to patch AI-found flaws. These stories highlight the growing impact of AI on jobs, security, and business operations.