OpenAI becomes latest AI firm to delay model rollout
OpenAI becomes latest AI firm to delay model rollout
Publish Date: 2026-06-29 15:52:00
Source Domain: foxbaltimore.com
Using an unordered list, summarize the following article with between 4 and 8 key points. WASHINGTON (TNND) — OpenAI became the latest high-profile artificial intelligence developer to restrict access on its newest models after concerns raised by the Trump administration over potential cybersecurity risks.The company said it is limiting access to its newest AI models, which operate under the name of GPT-5.6, are only initially being made available to a small group of trusted partners that have been approved by the Trump administration. It is the company’s most advanced AI model so far designed to perform more complex tasks than earlier versions.OpenAI said it hoped to make GPT-5.6 more broadly available in the coming weeks once the review is complete.“We don’t believe this kind of government access process should become the long-term default. It keeps the best tools from users, developers, enterprises, cyber defenders, and global partners who need them,” the company said in a blog post. “We are taking this short-term step because we believe it is the strongest path to broader availability in the coming weeks.”The restrictions are the second time this month that a leading American developer has delayed or limited access to a new model after concerns from the Trump administration. The moves are testing a new White House review process intended to prevent systems from being misused and raising questions about whether additional government scrutiny is slowing American companies in their race against China.The reviews come after President Donald Trump signed an executive order earlier this month giving the White House more oversight over the release of powerful new AI models. Under the order, AI companies are asked to voluntarily share new models with advanced hacking capabilities with the government up to 30 days, which allows federal agencies to discern what threats the products may present to financial systems, national security and other sensitive areas.The order is still in its early stages and there isn’t yet an established framework for companies to operate under as they try to compete with each other domestically and stay ahead of China. The government intervention has raised concerns among some industry experts who argue the review process doesn’t have clear standards and could become increasingly burdensome as models grow more powerful.“We need the government to be more specific about it have a very narrow and clearly defined process in terms of how that power is exerted,” said Adam Peruta, an associate professor and director of Syracuse University’s Advanced Media Management program. “But let’s face it, this is all new, different, uncharted territory. It’s not releasing a simple software update here and there.”Anthropic has spent the last two weeks working with the administration to get export controls lifted on its latest models over cybersecurity concerns. The Trump administration restricted access to Anthropic’s Fable 5 and Mythos 5 by foreign companies, governments and individuals.Fable and Mythos have been at least partially shut down for the last two weeks as the company tried to figure out how to address cybersecurity concerns raised by the Trump administration. Anthropic and other cybersecurity experts have disputed the validity of the security concerns, and some analysts have criticized the administration for picking winners and losers in the AI race.Administration officials are concerned that increasingly powerful AI systems could be used to perpetuate cyberattacks, develop malicious software or help foreign adversaries if they are released without safeguards.Most of the major American AI developers have agreed to share their models with the government for review by the Center for AI Standards and Innovation, a Commerce Department office created to evaluate advanced AI systems for potential national security and cyber threats.Prior to the executive order, the Trump administration had broadly taken a hands-off approach to regulating AI to maintain the United States’ competitive advantage as a global leader in the industry’s development. Trump has pushed back on congressional efforts to put guardrails in place for the industry despite mounting bipartisan concerns about the potential consequences of letting it mostly regulate itself.The decisions to delay the release of OpenAI and Anthropic’s newest models has raised similar concerns about more companies having to slow-walk the release of new models and losing an edge against China.“There’s always that chance that it’s going to slow down innovation, but what I’m more concerned with is what happens when China or another one of our adversaries creates a model that’s as powerful as what is under review right now, what are they going to do?” Peruta said.Some Chinese AI systems released recently have matched the performance of Mythos in some cybersecurity scenarios that also cost less to use and does not yet face restrictions in the U.S., raising the stakes in the global AI race. Many Chinese models are also open-source software, which allows anyone to use and modify it for free, while the most powerful American-made models come with a fee or are restricted to a limited set of users.
