In a significant development for AI governance, the White House has reportedly drafted guidance that would allow federal agencies to circumvent risk flags raised by Anthropic, a leading AI safety company. The move comes as the European Union's efforts to pass a watered-down version of its AI Act have collapsed, dealing a blow to regulatory harmonization.
According to sources, the draft guidance aims to streamline the approval process for AI models used by government agencies, bypassing Anthropic's internal risk assessment system. Anthropic, known for its emphasis on safety, had flagged several models as potentially harmful, but the White House argues that the current process is too slow for national security and economic competitiveness.
Meanwhile, EU lawmakers and member states failed to reach an agreement on a scaled-back version of the AI Act, which was intended to address concerns about overly strict regulations stifling innovation. The collapse has left the bloc without a unified framework for AI oversight, prompting criticism from both industry and advocacy groups.
In other AI news, AWS announced a trio of strategic updates, including deeper integration with OpenAI on Bedrock and a split of its Amazon Connect service. Booking Holdings reported that its AI agents have significantly reduced costs and boosted customer engagement, underscoring the growing impact of generative AI in travel tech.