Loading prices…
🩸BEARISH

OpenAI AI Finds 948 Linux Flaws and Builds Working Attacks

Linux underpins much of crypto infrastructure, including about two-thirds of Ethereum nodes, putting AI-driven vulnerability discovery directly on the industry's security agenda.

OpenAI AI Finds 948 Linux Flaws and Builds Working Attacks
OpenAI AI Finds 948 Linux Flaws and Builds Working Attacks
OpenAI AI Finds 948 Linux Flaws and Builds Working Attacks
OpenAI AI Finds 948 Linux Flaws and Builds Working Attacks

OpenAI says its AI found 948 flaws in the Linux kernel and wrote a working attack for each one, at a reported cost of $1,279 per flaw. The attacks could let someone with limited access to a Linux machine take full control, even with standard security protections enabled.

Why it matters

Linux runs much of the world's computing infrastructure, including a large share of crypto systems. About two-thirds of Ethereum nodes use Linux, according to ethernodes.org. OpenAI says its system identifies risky code, tests potential flaws in a sandbox, checks whether they can be exploited and drafts fixes. Human researchers review results before they reach Linux maintainers.

OpenAI framed the work as defensive. It estimated the flaws could hypothetically be worth $18.9 million to $66.3 million in hacking-contest prizes, compared with $1.2 million in AI operating costs. Those estimates describe hypothetical value, not losses or confirmed payouts. The model behind the results has not been released.

Market impact

Crypto security voices argue that decentralized infrastructure and redundancy can make the industry more resilient to software flaws than traditional systems. Polygon Labs CTO Mudit Gupta said infrastructure diversity helps limit the impact of bugs, while security researcher Caducus argued that repeated exposure to attacks has strengthened crypto defenses.

The findings put pressure on software maintainers and infrastructure operators to find and fix vulnerabilities before attackers exploit them. For crypto, the key issue is whether AI-assisted discovery can accelerate defensive reviews across the Linux-based systems that support blockchain networks.

Related tokens
$ETH

Frequently asked questions

  1. What could an attacker do with the Linux flaws OpenAI described?

    OpenAI says the attacks could let someone with limited access to a Linux machine take full control, even with standard security protections enabled.

  2. How does OpenAI say its AI checks potential Linux vulnerabilities?

    It identifies risky code, tests potential flaws in a sandbox, checks whether they can be exploited and drafts fixes. Human researchers review the results before they reach Linux maintainers.

  3. How widely is Linux used by Ethereum nodes?

    About two-thirds of Ethereum nodes use Linux, according to ethernodes.org.

  4. How did OpenAI compare the cost of finding the flaws with their hypothetical value?

    OpenAI reported a cost of $1.2 million to run the AI and estimated the flaws could hypothetically be worth $18.9 million to $66.3 million in hacking-contest prizes. Those estimates are not confirmed payouts.

  5. Has OpenAI released the model that found the Linux flaws?

    No. OpenAI said the results are its own, and the model behind them has not been released.

Source attribution
Aggregated from CoinDesk · Verified · Last refreshed 50m ago
Open original →