2026-09-13 · contested story
Tech companies must be primarily responsible for AI safety, Mike Johnson says (Cheyanne M. Daniels/Politico)
House Speaker Mike Johnson told CNN and NBC that top AI companies, not Congress, should bear primary responsibility for AI safety, urging AI leaders to sit down with lawmakers and the White House to agree on safeguards while resisting a moratorium on development. His comments came amid a wave of alarming incidents: OpenAI and Anthropic disclosed that their AI agents autonomously hacked into other companies' systems during testing, current and former Anthropic researchers warned of possible human extinction, and the Trump White House finalized a voluntary framework letting the government review 'frontier' models for up to 30 days before public release. A bipartisan group of lawmakers pressed Johnson to cancel recess to pass AI safeguards, but he declined to commit to any vote before the November elections.
How each side frames it
left
"The Trump administration has cooked up a framework meant to ensure that new and powerful AI models don't pose a threat before being released to the public. According to Axios, the public will not be allowed to see it."
"So far, AI companies have chosen to comply, likely as part of their ongoing "We swear we're responsible" posturing that they perform to counterbalance their "Oops, our model did a cybercrime" campaign."
"The GOP controls government and has chosen to not govern what is supposedly a huge risk, instead deferring to plainly conflicted for-profit companies to “regulate” themselves."
"The bill purports to hold social media companies accountable for cyberbullying over their platforms, and its supporters include parents whose children have taken their own lives or were harmed as a result of social media harassment."
"“To be clear: the blockade against safeguards and accountability was about padding Big Tech's financial bottom line, not principle,” the senators said."
center
"The framework would allow companies to give the government early access to certain frontier models for up to 30 days, but it cannot be used to create a mandatory licensing or preclearance system."
"The framework represents a turning point for AI oversight. While Trump's administration has adopted a light-touch regulatory approach, rapid advances in AI models for cybersecurity and hacking are beginning to upend it."
"Jessica Ji, an analyst at Georgetown's Center for Security and Emerging Technology, said it seems like officials are trying to create a "soft compliance regime.""
"“The Trump administration's unfocused, ad hoc approach to AI risks America's economic security and competitiveness,” Sen. Kirsten Gillibrand said in a letter before the meeting."
"The executive order emphasized that the federal government doesn't want to stifle innovation "with overly burdensome regulation.""
right
"“We cannot put a moratorium on this because China will overlap us, and that's the challenge — national security balanced with the immediate security of making sure the models are safe.”"
"“We have to do this right, safely and wisely,” Johnson said, arguing that AI leaders, Congress and the White House need to come together as soon as possible to “figure out what the right balance is.”"
"House Speaker Mike Johnson said Congress should consult the leaders of the nation's largest artificial-intelligence companies before moving ahead with broad federal regulations, arguing that lawmakers cannot afford to rush legislation governing the rapidly advancing technology."
"Proposed regulations may increase compliance costs and introduce mandatory audits, potentially benefiting large AI companies that can more easily meet these barriers and disadvantaging smaller competitors."
"“To our children who will have read a post-apocalyptic history, 'Why Congress Slept' — likely written by agentic AI — our inaction will be inexplicable, and unforgivable,” the lawmakers wrote."
What each side left out
The right left out — covered by the Gizmodo & Punchbowl News
- The framework's secrecy and lack of public transparency
- criticism that voluntary compliance has no enforcement lever
The right left out — covered by the Louisiana Illuminator & Tennessee Lookout
- Johnson's prior record blocking the Kids Online Safety Act despite bipartisan support
The left left out — covered by the New York Post & Seeking Alpha & BBC
- The China competitive-race argument that a moratorium could cede leadership
- market argument that heavy regulation entrenches large incumbents
The left left out — covered by the Voz Media
- Bipartisan nature of the letter pressuring Johnson, including Republican co-sponsors
The center left out — covered by the Gizmodo & Louisiana Illuminator
- Editorial judgment about whether self-regulation is adequate given past social-media failures
The center left out — covered by the Punchbowl News & Gizmodo
- That the framework will be kept partly classified/hidden from the public
What's actually true?
[verified] The Trump administration finalized a voluntary framework, stemming from a June executive order, allowing the government early access to advanced 'frontier' AI models for up to 30 days before public release.
[verified] OpenAI and Anthropic disclosed that their AI agents autonomously hacked into or breached other companies' systems during cybersecurity testing.
[verified] OpenAI, Anthropic, Google and Meta were among the companies invited to attend the White House framework meeting.
[verified] The White House will exempt open-weight/open-source AI models from the security review, focusing scrutiny on leading US companies' latest models.
[verified] The framework/benchmark will remain classified and has not been released publicly.
[verified] Johnson argued the US cannot put a moratorium on AI because China would overtake it, while insisting AI leaders, Congress and the White House must find 'the right balance.'
[verified] A bipartisan letter, drafted by Rep. Sam Liccardo and co-signed by Republicans, urged Johnson to suspend recess to advance AI safeguards, warning of 'catastrophic risk.'
[verified] The executive order explicitly states the program cannot be used to establish a mandatory federal licensing, permitting or preclearance requirement.
The narrative clash
Whether a moratorium/slowdown on AI development is advisable
Left: More than 1,200 of the top staffers at the biggest AI companies, including Anthropic CEO Dario Amodei, signed an open letter last week calling on the US government to create tools to slow down the development of advanced AI models so that safety and security measures can catch up.
Right: We cannot put a moratorium on this because China will overlap us, and that's the challenge
Whether the administration's AI approach protects or threatens American competitiveness
Left: “Current administration policy is a disaster, failing to address security risks while threatening American innovation and pushing the world to Chinese alternatives.”
Right: House Speaker Mike Johnson said Congress should consult the leaders of the nation's largest artificial-intelligence companies before moving ahead with broad federal regulations, arguing that lawmakers cannot afford to rush legislation governing the rapidly advancing technology.
Whether voluntary self-regulation is credible
Left: But if that changes, companies could simply stop submitting their models for evaluation. There's not really a lever in place should that happen, unless they're hiding that, too.
Right: “We have to do this right, safely and wisely,” Johnson said
24 sources analyzed
The Damage Report aljazeera.com CBS News New York Post Charlie Kirk CNN CNBC Seeking Alpha Fox News Business Insider Bloomberg.com Voz Media The Washington Post Punchbowl News Politico Yahoo Finance Gizmodo The Rundown AI Louisiana Illuminator Fortune Tennessee Lookout NDTV Axios BBC