{"version":"1.0","type":"rich","provider_name":"Acast","provider_url":"https://acast.com","height":250,"width":700,"html":"<iframe src=\"https://embed.acast.com/$/68a43f4573bf5b62987006aa/6a2a3738438158000bb79dcc?\" frameBorder=\"0\" width=\"700\" height=\"250\"></iframe>","title":"AI Gets Cheaper, Agents Start Paying, and Regulation Tightens","description":"In today’s episode of AI in Flow, Claire and Peter unpack a market-wide shift from benchmark bragging rights to hard business fundamentals. OpenAI is reportedly weighing significant token price cuts as enterprise buyers scrutinise costs and competition heats up. Visa’s collaboration with OpenAI pushes AI from assistant to actor, enabling agent-initiated purchases—raising new questions around controls, liability, fraud, and chargebacks. Anthropic reverses a policy that quietly degraded Claude for some frontier-development users and commits to making safety interventions visible, while also proposing tougher U.S. rules for frontier models, including pre-deployment testing and incident reporting. Plus: Oracle’s massive data-centre capex plans and what they signal about infrastructure constraints, ongoing market jitters around AI margins, a lawsuit alleging retaliation over Grok safety concerns, and Canada’s proposed Digital Safety Act adding pressure for safety-by-design and clearer labelling.","author_name":"Six & Flow"}