OpenAI expands access to powerful models to boost cybersecurity | Ukraine news

Source Domain: mezha.net

Author:

Using an unordered list, summarize the following article with between 4 and 8 key points.
OpenAI is opening its most capable models to verified government and business users. The move raises questions about oversight and competing industry strategies.

OpenAI is expanding access to its most powerful artificial intelligence models with the aim of strengthening cybersecurity at all levels – from federal agencies to state and local structures, as well as businesses. This decision differs significantly from Anthropic’s approach, which bets on more restricted access control to its models, considering it a key factor in enhancing global cybersecurity.
Such divergent positions in the field reflect a broad discussion in the world of artificial intelligence: technologies are evolving faster than legislation, regulatory frameworks, and societal norms, which leads to different risk management strategies.
Some companies choose the path of maximum speed in implementing innovations, while others approach the issue with greater caution, considering social consequences.
Previously, OpenAI’s “Cybersecurity Trusted Access” program was available only to selected partners. Now the company tells CNN that it is opening access to all verified levels of government – from federal agencies to state and local offices – for users who have passed verification and received approval, granting them access to special versions of OpenAI models with less restrictive parameters.
“We, as a company, do not believe that we should be the sole determinants of who gets access to our tools and what is the highest priority,” said Sasha Baker, head of national security policy at OpenAI, in an interview with CNN.
Against this backdrop, OpenAI seems to have a very different path from Mythos by Anthropic – a model that sparked concern due to the discovery and exploitation of software vulnerabilities.
Anthropic is building its strategy through Project Glasswing – a tightly controlled consortium that works closely with federal, regional, and local representatives.
The company believes that a slower and more cautious approach helps slow the AI arms race, which could be exploited by malicious actors.
Meanwhile, OpenAI previously granted access to its most powerful models to some companies and vetted independent security researchers. Now Baker has stated that the company aims to open the doors as wide as possible for users.

“We, as a company, do not believe that we should be the sole determinants of who gets access to our tools and what is the highest priority.”
– Sasha Baker

“We need to democratize our ability to raise cybersecurity for everyone who needs it, not leave it to Fortune 50 or the largest, most expensive companies that can afford it,” Baker said.

“No one should panic,” she said. “But this is a moment when we must act in a coordinated manner and do so with a certain effectiveness and urgency.”
– Sasha Baker

OpenAI also recently conducted a practical seminar in Washington with participants from various federal agencies – from the Pentagon to the White House, the Department of Homeland Security, and the Defense Advanced Research Projects Agency – to test the latest model and its capabilities in cybersecurity. The company plans to return to Washington in the coming weeks to gather feedback on tools and proposed policies.
“We will receive some guidance from the White House on where they want to push this issue and how they see the role of AI companies,” Baker said.
Representatives from OpenAI and other tech companies – namely Anthropic, Google, and financial institutions – were at the White House for a meeting with the National Cyber Director to discuss AI safety and cybersecurity.
OpenAI also unveiled a proposed “plan of action” for coordinating cybersecurity between government and private industry in the so-called Era of Intelligence. The company plans to roll out new security features for ChatGPT accounts and other tools soon to help users improve their own cyber hygiene.
Prospects and Regulatory Context
With this course of action, OpenAI aims to strengthen cybersecurity at all levels – from the federal government to regional structures and the private sector. At the same time, questions of access control, accountability, and regulatory oversight remain key topics for governments and the industry in the face of rising cyber threats.
In this context, the dialogue between openness and regulatory constraints could determine the future use of artificial intelligence and its impact on cybersecurity worldwide.