OpenAI released a letter Thursday warning that within months, artifical intelligence (AI) models could become powerful enough to allow bad actors to launch increasingly sophisticated cyberattacks on hospitals, water treatment plants and other critical infrastructure.
More than 100 other companies across banking, tech and cybersecurity signed the letter, including Accenture, Anthropic, Capital One, Google, Microsoft and Visa.
Hugging Face, an open-source AI model repository whose systems were hacked into by an OpenAI agent in July, also signed the letter.
“Today’s AI advances are already giving defenders new ways to fix weaknesses that have accumulated for years. If we act decisively, we can use the defenders’ window to make our digital world much more secure,” according to an excerpt of the letter.
DEM SENATOR PRESSES OPENAI, ANTHROPIC FOR ANSWERS IN AI HACKING PROBE
To fix the problem, OpenAI called on all organizations to make cybercrime defense “an immediate leadership priority.”
Cybersecurity organizations need to “lead the response to defend against sustained AI-enabled attacks, including testing defenses continuously against frontier cyber capabilities, strengthening existing tools with AI, and working with technology partners to close gaps now,” according to the letter.
Governments around the world were also given a call to action, with OpenAI saying that they should strengthen cyber defenses at the local, national and international levels.
“Give hospitals, water utilities, and local governments access to capable defensive AI, authorized testing, and hands-on support through trusted security providers and partners. Impose costs on attackers,” OpenAI told governmental bodies in the letter.
OPENAI LAUNCHES NEW CHATGPT EXPERIENCE WITH PARENTAL CONTROLS FOR TEENS
In a portion of the letter addressed to “frontier AI companies,” OpenAI said it should give critical-infrastructure operators access to powerful models, funding, training and technical help.
Operators should also build tools to track how AI agents are being used, test systems for vulnerabilities, fix problems responsibly, and share useful security information with governments and other defenders, OpenAI said.
There have already been a number of AI-enabled cyberattacks, perhaps most notably when Anthropic accused a Chinese state-sponsored group last September of manipulating its Claude Code tool in an attempt to hack into 30 global targets, allegedly breaching several.
“This marks the first documented case of agentic AI successfully obtaining access to confirmed high-value targets for intelligence collection,” according to an incident report from Anthropic.
Anthropic did not identify the victims of the cyberattack, but said major tech companies, financial institutions, chemical manufacturing companies and governments were among those targeted.
Anthropic has also been an industry leader in warning about how rapidly-advancing AI must be contained to avoid disastrous attacks on civilian infrastructure.
In April, the company announced Project Glasswing, an initiative to secure the world’s most critical software.
In the announcement, Anthropic said that one of its frontier models, Claude Mythos, had reached “a level of coding capability where they can surpass all but the most skilled humans at finding and exploiting software vulnerabilities.”
OpenAI and Anthropic did not immediately respond to FOX Business’ request for comment.
Read the full article here
