OpenAI, Anthropic and Google DeepMind have been holding talks for several weeks on the safety of advanced artificial intelligence systems, it was announced. OpenAI’s head of global policy, Chris Lehane, confirmed that the discussions are focused on safety testing, common standards and competition law; he is also in Washington to work with US lawmakers on large-scale risks. Anthropic CEO Dario Amodei called for industry cooperation to slow the pace of advanced model development when necessary and prevent risks. Sam Altman, Demis Hassabis and Elon Musk of SpaceXAI supported the call. Altman said independent third parties could conduct safety audits at OpenAI. The talks were said to have begun before Amodei’s article on Saturday. Executives had said that coordination could raise antitrust concerns. While Amodei proposed a narrow legal exemption, Lehane argued that it was unnecessary.
Why it matters
These discussions are putting efforts to address the safety of advanced artificial intelligence models through common tests and standards, rather than company-specific practices, on an institutional footing. This effort concerns not only the companies developing the models but also US lawmakers working on large-scale risks and independent third parties that could carry out potential audits. The call to slow the pace of development when necessary highlights the tension between safety goals and competition among companies. However, the parties do not agree on how coordination among rivals can be reconciled with competition law or whether a narrow legal exemption is needed. The main unresolved issue, therefore, is the legal framework under which common safety practices can be implemented without violating antitrust limits.
Background
OpenAI is not a new name in the FikirPilot archive: in the past 90 days, we have published 90 news stories mentioning the name; the latest is dated September 16, 2026.