Latest AI News

Ilya Sutskever, OpenAI co-founder, has been named CEO of Safe Superintelligence following Daniel Gross's departure to Meta. The $32 billion AI startup rejected Meta's acquisition offer earlier this year, emphasizing its commitment to independent development of safe AI systems. Co-founder Daniel Levy becomes president, maintaining leadership continuity. This executive shuffle occurs amid intensifying competition in the AI sector, with U.S. tech giants facing growing challenges from Chinese AI models gaining global traction. Sutskever aims to maintain the company's focus on technical excellence and safety in AI development.

The EU is developing a General-Purpose AI Code of Practice to help companies comply with the AI Act, scheduled to take effect in August 2025. Though the Code may not be ready until late 2025, it aims to provide guidance on transparency, copyright compliance, risk assessment, and governance structures for AI systems. The non-statutory document will serve as an interim guide until binding European standards are established. Companies following the Code may receive favorable compliance assessments, though adherence to the Code alone doesn't guarantee full compliance with the AI Act. The initiative involves collaboration between academic experts, industry representatives, and civil society members to ensure responsible AI development within the EU.

The EU's AI Act entered into force in August 2024, but faces implementation delays. Early obligations apply from February 2025, with GPAI requirements due August 2025. Poland has proposed postponing enforcement deadlines as technical standards remain unpublished. Industry faces uncertainty while awaiting clear guidance, with stakeholders debating between flexibility and immediate safeguards. Systems like ChatGPT are central to regulatory discussions as the EU works to establish what may become a global benchmark for AI regulation.

OpenAI has publicly disavowed any partnership with Robinhood regarding tokenized versions of its shares. Robinhood recently launched blockchain-based "stock tokens" for over 200 private companies on the Arbitrum blockchain, targeting European users. OpenAI emphasized that these tokens are not actual OpenAI equity, stating that any equity transfer requires their approval, which wasn't granted in this case.