AI bosses' safety push sparks rift inside OpenAI and Anthropic (www.ft.com)

Tensions are rising within OpenAI and Anthropic as their CEOs, Sam Altman and Dario Amodei, advocate for a deliberate slowdown in AI development to mitigate safety risks. Both companies are responding to increasing concerns from researchers about the potential dangers of advanced AI, with Amodei's recent essay calling for independent model evaluations and enhanced collaboration among AI labs and governments. However, despite this unified public stance, internal implementation remains vague, leading to apprehensions among employees about how these changes could impact their work and the security of their technology. Key proposals include allowing external evaluators access similar to employees, which some staff members fear could compromise sensitive systems in a highly competitive landscape. While efforts from independent organizations to assess AI safety are gaining traction, critics argue that the lack of meaningful oversight limits their effectiveness. The CEOs push for antitrust waivers to facilitate collaboration on safety measures, although the challenge remains to balance openness with protecting intellectual property. As discussions continue, the broader AI community is closely watching how these developments might shape future regulations and safety protocols in AI technology.