Artificial Intelligence
Bank of America assigns 4,000 staff to fix AI-found flaws
Bank of America said 1,000 contractors and 3,000 employees were working on software vulnerabilities identified by AI models, and Brian Moynihan said the bank spent a few hundred million dollars on the work and anti-money-laundering requirements.
- Bank of America said 1,000 outside contractors and 3,000 internal employees had been working on the software fixes since the second quarter.
- Brian Moynihan said the bank spent a few hundred million dollars in the third quarter on software work and anti-money-laundering requirements together.
- Anthropic released Mythos in April to selected partners through Project Glasswing and said the model had found thousands of serious security vulnerabilities.
- OpenAI followed in May with Daybreak.
- The article said Bank of America was one of the first major US companies to disclose how many people it had assigned to fixing vulnerabilities found by AI.
Bank of America said it had 1,000 contractors and 3,000 employees working on software vulnerabilities found by AI models. Brian Moynihan said the work had been under way since the second quarter, and that the bank had spent a few hundred million dollars in the third quarter on software work and anti-money-laundering requirements combined.
Anthropic released Mythos in April to selected partners through Project Glasswing. The company said the model had found thousands of serious security vulnerabilities.
OpenAI then followed in May with Daybreak. The article said the early warnings and countermeasures from the labs were first treated by critics as public relations.
It said security incidents such as the Hugging Face hack, together with independent evaluations, later changed the discussion. The article also said Bank of America was among the first large US companies to say how many people it had put on fixing AI-found vulnerabilities, and Moynihan expected those costs to fall over time.
Named in this story
People
- Brian Moynihan
- said contractors and employees were working on the software work and spending
Companies
- Bank of America
- said it had assigned staff to fix vulnerabilities found by AI
- Anthropic
- released Mythos and said it found thousands of serious vulnerabilities
- OpenAI
- followed Anthropic with its own cybersecurity programme
- Hugging Face
- its hack was cited as one of the incidents that changed the discussion
Organisations
- Project Glasswing
- the programme through which Anthropic released Mythos
Products and systems
- Mythos
- the model Anthropic released to selected partners
- Daybreak
- OpenAI’s cybersecurity programme launched in May
How the source tells it
The article is largely straight reporting, but it briefly uses a vindication narrative about AI security claims and the wider debate around them.
- vindication it contrasted early dismissal of the labs’ warnings with later claims that the debate had shifted