根据 Semafor 的审查结果，OpenAI 和 Anthropic 在今年第三季度都向立法者游说，讨论人工智能企业之间如何在不违反反垄断法规的前提下共享信息的问题。
Both OpenAI and Anthropic lobbied lawmakers on the ability of AI companies to share information with one another without running afoul of antitrust rules in the third quarter of this year, according to federal lobbying disclosures reviewed by Semafor.
Anthropic 的游说文件中提到，《对抗性威胁与安全风险合作法案》（Collaboration on Adversarial Threats and Security Risks Act）是该公司在国会山关注的重点议题。该法案旨在允许各研究机构在安全风险问题上展开合作；据 Semafor 早前的报道，这项立法可能会作为国防政策法案的一部分被正式通过。
Anthropic’s lobbying form lists the Collaboration on Adversarial Threats and Security Risks Act, a bill that would allow labs to coordinate with each other on security risks, as a point of focus for the company on Capitol Hill. The legislation could make it into law as part of must-pass defense policy legislation, Semafor previously reported.
OpenAI 的游说文件内容较为笼统，仅将“人工智能安全领域的合作与信息共享（包括拟议中的反垄断保护措施”列为在华盛顿讨论的议题。这些游说文件并未明确说明这些公司是支持还是反对相关提案。
OpenAI’s lobbying form is less specific, listing “AI-security collaboration and information-sharing, including proposed antitrust protections” as topics discussed in Washington. Lobbying forms don’t say whether companies are arguing for or against the proposals discussed in meetings.
鉴于上个月人工智能行业领导者提出的“协调减缓人工智能发展速度”的倡议，这类反垄断游说活动显得尤为重要。这两家公司经常就拟议中的人工智能监管法案、出口控制措施以及数据中心相关立法向华盛顿的立法者进行游说。
The antitrust lobbying is of interest given the talks around a coordinated AI development “slowdown” pitched by AI leaders last month. The companies regularly lobby Washington lawmakers on proposed AI regulatory bills, as well as legislation on export controls and data centers.
OpenAI 的全球事务负责人 Chris Lehane 认为，专门针对人工智能模型开发速度制定反垄断豁免条款并无必要。不过，特朗普政府的官员批评这些人工智能领导者要求政府给予此类豁免，称他们完全有能力自行决定减缓开发速度。
OpenAI’s head of global affairs, Chris Lehane, has argued that an antitrust exemption specifically for model pacing would not be necessary. But Trump administration officials criticized AI leaders for calling for such a carve-out, saying they were capable of deciding to slow down among themselves without the government’s blessing.
根据参议院的游说文件显示，随着人工智能竞争的加剧以及高调的安全事件对华盛顿的讨论产生压力，Meta 在 2026 年第三季度向华盛顿立法者投入了 590 万美元用于游说活动——这是当时所有大型科技公司中支出最多的。
Meta spent $5.9 million lobbying Washington lawmakers in the third quarter of 2026 — the most of any big tech firm during that period — as the AI race grew more competitive and high-profile hacks put pressure on the AI safety debate in Washington, according to Senate lobbying filings.
Meta 通常是大型科技和 AI 同行中游说开支最大的公司，但 OpenAI 和 Anthropic 的季度开支不断增加，在上半年分别几乎翻倍和翻三倍，因为它们正在应对华盛顿关于遏制其技术的讨论。
Meta often is the biggest lobbying spender among its Big Tech and AI counterparts, but OpenAI and Anthropic have been spending more and more each quarter, nearly doubling and tripling their spends in the first half of the year, respectively, as they navigate Washington’s discussions about containing their technology.
Google、Amazon 和 Apple 的开支落后于 Meta，但超过了 AI 和芯片公司，尽管 Nvidia 也创下了单季度 220 万美元的历史新高，因为这家芯片制造商在特朗普政府中保持主导地位和影响力。这三家公司都在一系列科技政策议题上建立了庞大的游说组合，包括儿童网络安全、移民、税收、教育和能源政策。
Google, Amazon, and Apple lagged behind Meta in spending but beat AI and chip companies, though Nvidia also posted its highest-ever quarter at $2.2 million as the chipmaker stays dominant and influential in the Trump administration. All three companies have built up massive lobbying portfolios on a range of tech policy topics including kids’ online safety, immigration, taxes, education, and energy policy.