ToS

Anthropic’s claim that rivals prompted Claude more than 16 million times has made ToS enforcement a focal point in the latest AI-competition coverage.

Who they are

ToS, or terms of service, appears in coverage as the contractual rulebook through which technology platforms define acceptable use, data rights, moderation powers, and access to services. The stories treat it less as a standalone company than as the mechanism invoked by platforms including Facebook, YouTube, X, Mozilla, Bluesky, Zoom, and Anthropic when disputes emerge.

The recent arc

Recent coverage peaked in 2024Q2, led by YouTube’s strengthened enforcement against third-party ad-blocking apps that it said violated its ToS. By 2024Q4 and into 2025, the focus broadened from platform-access enforcement to the scope and legitimacy of policy changes: Mozilla faced backlash over Firefox terms critics called overly broad, while Bluesky revised policies and Community Guidelines around the EU Digital Services Act, the UK Online Safety Act, and the US Take It Down Act.

The latest phase is centered on AI rivalry and control of model access. Anthropic revoked OpenAI’s Claude API access in August 2025 over alleged ToS violations, and in February 2026 accused DeepSeek, MiniMax, and Moonshot of using more than 16 million Claude prompts and distillation to train competing products. X’s update asserting rights to the Twitter trademark shows that terms also remain a vehicle for platform control of brands and legal positioning.

The tension

Coverage repeatedly circles the gap between platforms’ need to enforce rules and users’, developers’, and competitors’ objections to those rules. YouTube’s action against ad-blocking apps pits service control against alternative clients; Mozilla’s revision raised fears about data use; and Anthropic’s allegations against OpenAI, DeepSeek, MiniMax, and Moonshot put API terms at the center of whether model evaluation or extraction becomes impermissible competitive copying.

Why it matters

If this trajectory holds, terms of service will increasingly operate as an enforcement layer alongside product design, regulation, and litigation, especially where platforms can restrict API access or define use of data and intellectual property. The unresolved question in the coverage is whether such rules will be seen as proportionate safeguards against misuse or as opaque tools that concentrate platform power; regulatory scrutiny and user backlash make that distinction consequential.

Related: Anthropic says DeepSeek, MiniMax, and Moonshot violated its ToS by pro · YouTube strengthens its “enforcement on third-party apps” violating it · Zoom updates its terms to let the company train its AI models using so · Facebook · Bluesky · YouTube