By Paul Sandle
LONDON, May 12 (Reuters) - U.S. artificial intelligence giant OpenAI said it was granting access to its latest models including GPT-5.5-Cyber to Deutsche Telekom, BBVA and dozens more European companies to help bolster their resilience to vulnerabilities in their systems.
OpenAI's "Trusted Access for Cyber" programme gives verified companies in vital sectors such as financial services, telecoms, energy and public services access to its models, including precise safeguards for defensive
work.
OpenAI's MD for EMEA, Emmanuel Marill, said there was an important balance to be struck between access, usefulness and safety as AI became more capable.
"We need to block dangerous activity, while making sure trusted defenders have tools that are genuinely useful in protecting systems, finding vulnerabilities and responding to threats quickly," he said on Tuesday.
The release of Mythos by OpenAI's rival Anthropic last month significantly upped the risks posed to banks and other companies from new frontier AI models.
Their capabilities to code at a high level have given them an unprecedented ability to identify cybersecurity risks and devise ways to exploit them, raising fears they could be used to destabilise banks and other companies.
OpenAI has offered the European Commission open access to cybersecurity features, Brussels said on Monday, but it added that Anthropic had not been as forthcoming.
Former British finance minister George Osborne, who heads the company's "OpenAI for Countries" initiative, on Monday sent an explanatory letter to the Commission, saying that democratizing access to defensive tools could strengthen shared security, support public safety and reflect European priorities.
OpenAI also said on Monday it was setting up a new company with more than $4 billion in initial investment to help organisations build and deploy AI systems, and would acquire AI consulting firm Tomoro to quickly scale up the unit.
(Reporting by Paul Sandle. Editing by Mark Potter)
