OpenAI等四大巨头采纳新AI抓取退出标准

OpenAI, Google, Microsoft, Anthropic adopt new web opt-out standard for AI training

Major AI labs have jointly pledged to honor an updated robots.txt-based protocol that lets publishers explicitly disallow AI crawlers from harvesting content for model training. While voluntary and not legally binding, this signals growing industry alignment on data provenance and publisher rights—especially amid ongoing copyright litigation. For early-stage AI application builders, this raises the bar for data sourcing: teams relying on unlicensed web scraping face higher compliance risk and potential model degradation as key domains opt out. Conversely, startups with proprietary or community-curated datasets gain leverage to offer licensed, domain-specific fine-tuning services—shifting value from generic inference to trusted, auditable data-model bundles.