OpenAI partners with Cathedral to find AI security flaws, raises bug bounty payouts

OpenAI has announced a partnership with Cathedral to find and fix security flaws in its AI systems more quickly. The company is also revising its bug bounty scheme, and payouts for critical reports are set to rise more than four times. OpenAI has not said what the new amounts will be.

The collaboration brings external security researchers into OpenAI's work on vulnerabilities. The bug bounty change is aimed at encouraging outside experts to report serious flaws. The details come from the company's own announcement, and the figures it gave are limited.

Building on an internal effort

The announcement follows an earlier internal exercise. A quarter of OpenAI's product engineers used Codex, the company's coding tool, to investigate security issues. The company is pairing that internal work with help from outside researchers.

OpenAI has also committed $1 billion, roughly Rs 9,500 crore, in discounted access for frontline cyber defenders over six months. The money is meant to give people working on cyber defence access to its tools at a lower cost.

What it means for readers in India

For Indian security researchers, the higher bug bounty payouts could make OpenAI's programme more attractive to work on. Anyone who finds a critical flaw in OpenAI's products may now have a stronger reason to report it through the programme. The company has not yet said when the new payout amounts will take effect.

OpenAI has not disclosed the revised bounty amounts, so the size of the increase in rupees is not known. The details of how the Cathedral partnership will work are also not yet clear.