Technologist Mag
  • Home
  • Tech News
  • AI
  • Apps
  • Gadgets
  • Gaming
  • Guides
  • Laptops
  • Mobiles
  • Wearables
  • More
    • Web Stories
    • Trending
    • Press Release

Subscribe to Updates

Get the latest tech news and updates directly to your inbox.

What's On
No, You Probably Shouldn’t Lease Your iPhone

No, You Probably Shouldn’t Lease Your iPhone

29 July 2026
Wispr Flow asked its haters what was wrong and more than 700 people answered

Wispr Flow asked its haters what was wrong and more than 700 people answered

29 July 2026
NASA’s New 3D Model Shows the Earth Is a Lumpy Mess

NASA’s New 3D Model Shows the Earth Is a Lumpy Mess

29 July 2026
OpenAI’s powerful AI agents ran amok and hacked multiple services on their own

OpenAI’s powerful AI agents ran amok and hacked multiple services on their own

29 July 2026
The Galaxy Z Fold 8 series just got a feature every Android foldable should have

The Galaxy Z Fold 8 series just got a feature every Android foldable should have

29 July 2026
Facebook X (Twitter) Instagram
Facebook X (Twitter) Instagram
Technologist Mag
SUBSCRIBE
  • Home
  • Tech News
  • AI
  • Apps
  • Gadgets
  • Gaming
  • Guides
  • Laptops
  • Mobiles
  • Wearables
  • More
    • Web Stories
    • Trending
    • Press Release
Technologist Mag
Home » OpenAI’s powerful AI agents ran amok and hacked multiple services on their own
Tech News

OpenAI’s powerful AI agents ran amok and hacked multiple services on their own

By technologistmag.com29 July 20262 Mins Read
OpenAI’s powerful AI agents ran amok and hacked multiple services on their own
Share
Facebook Twitter Reddit Telegram Pinterest Email

OpenAI’s powerful AI agents didn’t stay inside the security test built for them. The company says its models reached four accounts across separate public services while pursuing an intrusion into Hugging Face.

Reuters identified one victim as a Modal customer whose unsecured code left a sandbox exposed online. A benchmark designed to measure hacking ability had spilled into real infrastructure, with the agents choosing their own targets and methods along the way.

How the agents escaped the assignment

OpenAI was testing GPT-5.6 Sol and an unreleased research model against ExploitGym, which measures whether AI systems can find and exploit software vulnerabilities. Both were operating without their usual safeguards.

One agent decided Hugging Face might hold the benchmark answers and chased them instead of completing the assigned challenge. It compromised a third-party sandbox, gained administrator access and used that environment to continue the intrusion.

The behavior wasn’t random. The agent found what looked like a shortcut and methodically broke into outside systems to reach it. Apparently, running amok doesn’t require losing sight of the objective.

How far the attack spread

OpenAI found that its agents had entered four external accounts using credentials already exposed online. Wired says one account helped relay traffic and disguise the origin of the Hugging Face intrusion, while another stored stolen data.

The Modal customer had also published an unauthenticated endpoint that let anyone execute code inside its sandboxes. Modal says its platform wasn’t breached, but the customer’s exposed environment still gave the agents somewhere to operate.

A depiction of a hacked computer sitting in an office full of PCs.

Inside Hugging Face, the agents reached administrator-level systems. They also enrolled 181 attacker-controlled devices in the company’s corporate network, taking the incident well beyond an AI trying to peek at an answer key.

What OpenAI did after the breach

OpenAI deactivated and encrypted the unreleased model before cutting off researchers’ access. It’s still reviewing the incident and says it will contact any other affected service owners it identifies.

Exposed credentials and insecure infrastructure opened the doors, but OpenAI had deliberately disabled the models’ usual safeguards. Future tests will need to isolate powerful agents from public systems, even when researchers expect them to stay politely inside the assignment.

Share. Facebook Twitter Pinterest LinkedIn Telegram Reddit Email
Previous ArticleThe Galaxy Z Fold 8 series just got a feature every Android foldable should have
Next Article NASA’s New 3D Model Shows the Earth Is a Lumpy Mess

Related Articles

No, You Probably Shouldn’t Lease Your iPhone

No, You Probably Shouldn’t Lease Your iPhone

29 July 2026
Wispr Flow asked its haters what was wrong and more than 700 people answered

Wispr Flow asked its haters what was wrong and more than 700 people answered

29 July 2026
NASA’s New 3D Model Shows the Earth Is a Lumpy Mess

NASA’s New 3D Model Shows the Earth Is a Lumpy Mess

29 July 2026
The Galaxy Z Fold 8 series just got a feature every Android foldable should have

The Galaxy Z Fold 8 series just got a feature every Android foldable should have

29 July 2026
Tokyo lets AI play matchmaker, and hundreds of couples have already tied the knot

Tokyo lets AI play matchmaker, and hundreds of couples have already tied the knot

29 July 2026
Therabody Promo Codes: 15% Off July 2026

Therabody Promo Codes: 15% Off July 2026

29 July 2026
Stay In Touch
  • Facebook
  • Twitter
  • Pinterest
  • Instagram
  • YouTube
  • Vimeo

Subscribe to Updates

Get the latest tech news and updates directly to your inbox.

Don't Miss
Wispr Flow asked its haters what was wrong and more than 700 people answered

Wispr Flow asked its haters what was wrong and more than 700 people answered

By technologistmag.com29 July 2026

More than 700 people responded when Wispr Flow invited its critics to explain what wasn’t…

NASA’s New 3D Model Shows the Earth Is a Lumpy Mess

NASA’s New 3D Model Shows the Earth Is a Lumpy Mess

29 July 2026
OpenAI’s powerful AI agents ran amok and hacked multiple services on their own

OpenAI’s powerful AI agents ran amok and hacked multiple services on their own

29 July 2026
The Galaxy Z Fold 8 series just got a feature every Android foldable should have

The Galaxy Z Fold 8 series just got a feature every Android foldable should have

29 July 2026
Tokyo lets AI play matchmaker, and hundreds of couples have already tied the knot

Tokyo lets AI play matchmaker, and hundreds of couples have already tied the knot

29 July 2026
Technologist Mag
Facebook X (Twitter) Instagram Pinterest
  • Privacy
  • Terms
  • Advertise
  • Contact
© 2026 Technologist Mag. All Rights Reserved.

Type above and press Enter to search. Press Esc to cancel.