That company, per a report Monday (Nov. 16) by The Information, makes technology designed to determine why artificial intelligence (AI) models behave how they do, and how to keep them from going rogue.
At one point, a senior OpenAI executive came up with the notion of either acquiring Goodfire—valued at $1.25 billion earlier this year—for $5 billion or establishing some other sort of partnership, two sources familiar with the proposal told The Information.
While no deal was reached, the report contends that the negotiations underscore how a “once-obscure” corner of the industry is receiving newfound attention as AI-driven security breaches surge, and Anthropic and OpenAI worry about their companies’ ability to control their tech.
One of the sources said the deal didn’t come together partially because Goodfire executives think the startup will become more valuable than $5 billion if it stays independent and continues to sell technology to multiple AI developers, according to one of the people.
According to the report, Goodfire sees its software as especially useful as leaders at OpenAI, Anthropic and Google have admitted that traditional forms of tracking AI behavior are becoming less reliable.
The company’s approach involves reading their AI models’ “chains of thought”—the text that models generate as they reason through problems—and closely observing the actions the AI actually takes, the report said.
Writing about AI-powered cyberattacks Monday, PYMNTS noted that artificial intelligence-assisted frameworks can automate much of the work that once required skilled operators: move between systems, interpret results and decide which vulnerabilities to exploit next.
“That changes the cost equation for defenders,” the report said. “A company might spend months securing its most critical applications, only to discover that attackers have found a less-protected employee portal, supplier integration or legacy service.”
The PYMNTS Intelligence report “Scale Amplification: How Revenue Amplifies Agent-Driven Identity” found that large enterprises, with their bigger digital footprints, can be more vulnerable to today’s new AI-powered attacks employed by adversarial fraudsters.
“The asymmetry is becoming a challenge for enterprise CISOs and CFOs,” the report added. “Attackers can scale their operations through software, while defenders remain constrained by staffing, budgets and organizational complexity.”
PYMNTS Intelligence’s “2025 State of Fraud and Financial Crime in the United States” found that 68% of financial institutions had upped their cyber budgets in the last year.
For all PYMNTS AI and digital transformation coverage, subscribe to the daily AI and Digital Transformation Newsletters.