OpenAI confirms joint safety efforts with Anthropic, Google DeepMind days after Sam Altman agrees with Dario Amodei’s 3800-word AI slowdown essay
OpenAI has confirmed that it is working with competitors Anthropic and Google DeepMind on artificial intelligence (AI) safety measures. The joint initiative aims to confront mounting alarm among lawmakers and the public that uncontrolled computing models present grave economic hazards and national security liabilities, reports Bloomberg. Speaking during a Washington press briefing on Tuesday (September 15), OpenAI global policy chief Chris Lehane revealed that collaborative talks between the three frontier developers have already been actively progressing for multiple weeks. Earlier this week, a report by CNN cited sources as saying that the three companies are working together on an AI safety effort.Addressing regulatory questions surrounding the collaboration, Lehane emphasised that the tech firms do not require an antitrust exemption to exchange insights and align on security practices. Lehane told attendees: “It’s better to try to work together to prioritize safety.”
AI development pace debate after Anthropic CEO Dario Amodei write letter
Anxiety over catastrophic risks reached a boiling point over the weekend following a 3,800-word open letter by Anthropic CEO Dario Amodei, who called on development labs to deliberately dial back the speed of leading-edge model upgrades so computer scientists have adequate time to map and defuse emergent hazards. The plea won swift backing from high-profile technology figures, with OpenAI’s Sam Altman agreeing with the premise and xAI head Elon Musk voicing his support.Defending the partnership between market rivals, Lehane pointed out that competitor cooperation on consumer welfare has deep historical roots, comparing the dynamic to commercial aviation.“There are many instances over time where companies are trying to help each other on safety,” he said.
OpenAI courts lawmakers on governance bills
Lehane noted that his presence in the capital involved direct negotiations with American legislators, asserting that OpenAI is ready to endorse bipartisan legislation designed to tackle catastrophic technology scenarios.The specific legislative packages favoured by OpenAI include a federal regulatory blueprint crafted by Republican Representative Jay Obernolte and Democratic Representative Lori Trahan as well as emerging bipartisan Senate legislation led by Majority Leader John Thune, Republican Ted Cruz and Democrat Amy Klobuchar, which focuses squarely on heading off catastrophic computational harms.
