Bank of America hires 1,000 contractors to patch software flaws from Anthropic's Mythos

by@CNBC

Summary

Bank of America has hired 1,000 contractors to address software vulnerabilities exposed by Anthropic's advanced artificial intelligence model, Mythos, as part of a broader effort to bolster its cybersecurity amidst growing risks from AI. CEO Brian Moynihan stated during a CNBC interview that the bank undertook this action after the release of Mythos earlier this year, recognizing the urgent need to patch a wide range of internal and vendor code vulnerabilities. This initiative has led to significant financial outlays, with the bank spending "a couple hundred million plus" on software remediation and regulatory compliance, including anti-money laundering measures, in the third quarter. Moynihan also mentioned that while AI can offer productivity improvements, its ability to identify long-standing software flaws presents ongoing challenges for firms, necessitating a delicate balance between leveraging AI's capabilities and protecting against its risks.

Tokens

$BAC

Analysis

Anthropic: Anthropic is an AI company developing frontier models designed for safety and capability. It released Mythos in April to a select group of banks, tech firms, and government agencies, where the model demonstrated the ability to identify and exploit software flaws. This has prompted organizations like Bank of America to accelerate remediation work. Brian Moynihan: Brian Moynihan serves as chairman and CEO of Bank of America. He addressed the company's response to AI-driven security risks, including contractor hiring and remediation spending, during post-earnings interviews. Moynihan also noted the temporary nature of increased costs while emphasizing ongoing AI productivity initiatives. Bank of America: Bank of America is a major U.S. financial institution offering banking, investment, and wealth management services. In this news, the bank mobilized significant resources including external contractors to address software vulnerabilities exposed by advanced AI models. CEO Brian Moynihan highlighted these efforts during the third-quarter earnings discussion. Dual Use of AI: Organizations are balancing AI's productivity benefits with the need to counter its use in exposing system weaknesses. AI Security Impact: Frontier AI models are capable of discovering decades-old software vulnerabilities, leading major corporations to undertake rapid defensive patching efforts.

Categories

aimacrotechpolitics
View Original Tweet