The White House has shifted its discussions with Anthropic towards establishing specific rules for artificial intelligence security, marking a transition from general safety concerns to concrete regulatory frameworks. This development highlights the increasing priority of formalizing governance protocols within the rapidly evolving AI sector. The engagement aims to address potential risks associated with advanced AI systems through defined standards rather than voluntary measures.

The move comes as regulatory bodies intensify their examination of AI technologies. By focusing on rule-setting, the administration seeks to create a structured environment for AI development and deployment. This shift suggests that previous dialogues regarding safety protocols have evolved into actionable policy discussions.

Regulatory Focus

The transition to setting rules indicates a significant step in the government's approach to AI oversight. Key areas of focus are expected to include:

• Security Standards: Defining technical requirements for AI system robustness.
• Risk Management: Establishing protocols for identifying and mitigating potential threats.
• Compliance Mechanisms: Creating frameworks for enforcing adherence to new regulations.

The outcome of these discussions could set a precedent for future interactions between the government and AI companies. As the sector continues to advance, the establishment of clear security rules will likely play a critical role in shaping the trajectory of AI innovation and its integration into society.

Anthropic faces a class-action lawsuit alleging its Claude Max subscriptions provide far less access than advertised to users. Plaintiff Karl Kahn, a Washington, D.C., resident, claims the effective caps for the Max 5x and Max 20x options fall below the usage levels Anthropic markets on its website and other materials. The lawsuit seeks to represent U.S. consumers who purchased Max subscriptions since the tiers were introduced last year.

Anthropic offers three paid plans for individuals, including Claude Pro starting at $17 per month. According to its website, the Pro plan "offers at least five times the usage per session compared to our free service" during peak hours, with message limits varying based on length, attachments, and model features. The company later introduced Max plans priced at $100 and $200 per month, advertised to provide up to 5x and 20x the Pro plan's limits respectively.

Kahn argues in the complaint that "the actual usage provided by the Max 5x and Max 20x plans is far below the advertised amount of usage." After upgrading to the Max 20x plan in April, he reports encountering weekly ceilings quickly, including a five-hour period that consumed 15% of his weekly allotment. The filing highlights a gap between marketed capabilities and actual user experience.

Usage for Claude Code was constrained by weekly limits in July after some customers ran the tool "continuously in the background, 24/7." The underlying metric for usage is "tokens," the units large language models process to handle prompts and generate outputs. Costs can fluctuate significantly based on prompt complexity, complicating the application of standard subscription models to compute-intensive AI services.

Anthropic declined to comment on the lawsuit when contacted by email. The case raises questions about transparency in subscription models for AI services, where variable costs and technical constraints may conflict with flat-rate pricing expectations.