OpenAI’s latest AI models have a new safeguard to prevent biorisks

23/05/2025

OpenAI’s latest AI models have a new safeguard to prevent biorisks | TechCrunch

===============
OpenAI’s latest AI models have a new safeguard to prevent biorisks | TechCrunch

===============
[Skip to content](https://techcrunch.com/2025/04/16/openais-latest-ai-models-have-a-new-safeguard-to-prevent-biorisks/#wp–skip-link–target)

[![Image 2](https://techcrunch.com/wp-content/uploads/2024/09/tc-lockup.svg)TechCrunch Desktop Logo](https://techcrunch.com/)[![Image 3](https://techcrunch.com/wp-content/uploads/2024/09/tc-logo-mobile.svg)TechCrunch Mobile Logo](https://techcrunch.com/)

* [Latest](https://techcrunch.com/latest/)
* [Startups](https://techcrunch.com/category/startups/)
* [Venture](https://techcrunch.com/category/venture/)
* [Apple](https://techcrunch.com/tag/apple/)
* [Security](https://techcrunch.com/category/security/)
* [AI](https://techcrunch.com/category/artificial-intelligence/)
* [Apps](https://techcrunch.com/category/apps/)
* [Google I/O 2025](https://techcrunch.com/storyline/google-i-o-2025-live-coverage-gemini-android-16-updates-and-more/)

* [Events](https://techcrunch.com/events/)
* [Podcasts](https://techcrunch.com/podcasts/)
* [Newsletters](https://techcrunch.com/newsletters/)

Submit

Site Search Toggle

Mega Menu Toggle

### Topics

[Latest](https://techcrunch.com/latest/)

[AI](https://techcrunch.com/category/artificial-intelligence/)

[Amazon](https://techcrunch.com/tag/amazon/)

[Apps](https://techcrunch.com/category/apps/)

[Biotech & Health](https://techcrunch.com/category/biotech-health/)

[Climate](https://techcrunch.com/category/climate/)

[Cloud Computing](https://techcrunch.com/tag/cloud-computing/)

[Commerce](https://techcrunch.com/category/commerce/)

[Crypto](https://techcrunch.com/category/cryptocurrency/)

[Enterprise](https://techcrunch.com/category/enterprise/)

[EVs](https://techcrunch.com/tag/evs/)

[Fintech](https://techcrunch.com/category/fintech/)

[Fundraising](https://techcrunch.com/category/fundraising/)

[Gadgets](https://techcrunch.com/category/gadgets/)

[Gaming](https://techcrunch.com/category/gaming/)

[Google](https://techcrunch.com/tag/google/)

[Government & Policy](https://techcrunch.com/category/government-policy/)

[Hardware](https://techcrunch.com/category/hardware/)

[Instagram](https://techcrunch.com/tag/instagram/)

[Layoffs](https://techcrunch.com/tag/layoffs/)

[Media & Entertainment](https://techcrunch.com/category/media-entertainment/)

[Meta](https://techcrunch.com/tag/meta/)

[Microsoft](https://techcrunch.com/tag/microsoft/)

[Privacy](https://techcrunch.com/category/privacy/)

[Robotics](https://techcrunch.com/category/robotics/)

[Security](https://techcrunch.com/category/security/)

[Social](https://techcrunch.com/category/social/)

[Space](https://techcrunch.com/category/space/)

[Startups](https://techcrunch.com/category/startups/)

[TikTok](https://techcrunch.com/tag/tiktok/)

[Transportation](https://techcrunch.com/category/transportation/)

[Venture](https://techcrunch.com/category/venture/)

### More from TechCrunch

[Events](https://techcrunch.com/events/)

[Startup Battlefield](https://techcrunch.com/startup-battlefield/)

[StrictlyVC](https://strictlyvc.com/)

[Newsletters](https://techcrunch.com/newsletters/)

[Podcasts](https://techcrunch.com/podcasts/)

[Videos](https://techcrunch.com/video/)

[Partner Content](https://techcrunch.com/sponsored/)

[TechCrunch Brand Studio](https://techcrunch.com/brand-studio/)

[Crunchboard](https://www.crunchboard.com/)

[Contact Us](https://techcrunch.com/contact-us/)

![Image 4: The ChatGPT logo appears on a smartphone screen](https://techcrunch.com/wp-content/uploads/2025/01/GettyImages-2191707579.jpg?w=1024)

**Image Credits:**Jaque Silva/NurPhoto / Getty Images

[AI](https://techcrunch.com/category/artificial-intelligence/)

[](https://www.facebook.com/sharer.php?u=https%3A%2F%2Ftechcrunch.com%2F2025%2F04%2F16%2Fopenais-latest-ai-models-have-a-new-safeguard-to-prevent-biorisks%2F)[](https://twitter.com/intent/tweet?url=https%3A%2F%2Ftechcrunch.com%2F2025%2F04%2F16%2Fopenais-latest-ai-models-have-a-new-safeguard-to-prevent-biorisks%2F&text=OpenAI%E2%80%99s+latest+AI+models+have+a+new+safeguard+to+prevent+biorisks&via=techcrunch)[](https://www.linkedin.com/shareArticle?url=https%3A%2F%2Ftechcrunch.com%2F2025%2F04%2F16%2Fopenais-latest-ai-models-have-a-new-safeguard-to-prevent-biorisks%2F&title=OpenAI%E2%80%99s+latest+AI+models+have+a+new+safeguard+to+prevent+biorisks&summary=OpenAI+says+that+it+deployed+a+new+system+to+monitor+its+latest+AI+reasoning+models%2C+o3+and+o4-mini%2C+for+prompts+related+to+biological+and+chemical+threats.+The+system+aims+to+prevent+the+models+from+offering+advice+that+could+instruct+someone+on+carrying+out+potentially+harmful+attacks%2C+according+to+OpenAI%E2%80%99s+safety+report.+O3+and+o4-mini+represent+%5B%E2%80%A6%5D&mini=1&source=TechCrunch)[](https://www.reddit.com/submit?url=https%3A%2F%2Ftechcrunch.com%2F2025%2F04%2F16%2Fopenais-latest-ai-models-have-a-new-safeguard-to-prevent-biorisks%2F&title=OpenAI%E2%80%99s+latest+AI+models+have+a+new+safeguard+to+prevent+biorisks)[](mailto:?subject=OpenAI%E2%80%99s+latest+AI+models+have+a+new+safeguard+to+prevent+biorisks&body=Article%3A+https%3A%2F%2Ftechcrunch.com%2F2025%2F04%2F16%2Fopenais-latest-ai-models-have-a-new-safeguard-to-prevent-biorisks%2F)[](https://techcrunch.com/2025/04/16/openais-latest-ai-models-have-a-new-safeguard-to-prevent-biorisks/)

OpenAI’s latest AI models have a new safeguard to prevent biorisks
==================================================================

[Maxwell Zeff](https://techcrunch.com/author/maxwell-zeff/)

2:12 PM PDT · April 16, 2025

OpenAI says that it deployed a new system to monitor its latest AI reasoning models, [o3 and o4-mini](https://techcrunch.com/2025/04/16/openai-launches-a-pair-of-ai-reasoning-models-o3-and-o4-mini/), for prompts related to biological and chemical threats. The system aims to prevent the models from offering advice that could instruct someone on carrying out potentially harmful attacks, [according to OpenAI’s safety report](https://cdn.openai.com/pdf/2221c875-02dc-4789-800b-e7758f3722c1/o3-and-o4-mini-system-card.pdf).

O3 and o4-mini represent a meaningful capability increase over OpenAI’s previous models, the company says, and thus pose new risks in the hands of bad actors. According to OpenAI’s internal benchmarks, o3 is more skilled at answering questions around creating certain types of biological threats in particular. For this reason — and to mitigate other risks — OpenAI created the new monitoring system, which the company describes as a “safety-focused reasoning monitor.”

The monitor, custom-trained to reason about OpenAI’s content policies, runs on top of o3 and o4-mini. It’s designed to identify prompts related to biological and chemical risk and instruct the models to refuse to offer advice on those topics.

To establish a baseline, OpenAI had red teamers spend around 1,000 hours flagging “unsafe” biorisk-related conversations from o3 and o4-mini. During a test in which OpenAI simulated the “blocking logic” of its safety monitor, the models declined to respond to risky prompts 98.7% of the time, according to OpenAI.

OpenAI acknowledges that its test didn’t account for people who might try new prompts after getting blocked by the monitor, which is why the company says it’ll continue to rely in part on human monitoring.

O3 and o4-mini don’t cross OpenAI’s “high risk” threshold for biorisks, according to the company. However, compared to o1 and GPT-4, OpenAI says that early versions of o3 and o4-mini proved more helpful at answering questions around developing biological weapons.

![Image 5](https://techcrunch.com/wp-content/uploads/2025/04/Screenshot-2025-04-16-at-2.09.10PM.png?w=680)

Chart from o3 and o4-mini’s system card (Screenshot: OpenAI)

The company is actively tracking how its models could make it easier for malicious users to develop chemical and biological threats, according to OpenAI’s recently updated [Preparedness Framework](https://cdn.openai.com/pdf/18a02b5d-6b67-4cec-ab64-68cdfbddebcd/preparedness-framework-v2.pdf).

Techcrunch event

### Join us at TechCrunch Sessions: AI

#### Secure your spot for our leading AI industry event with speakers from OpenAI, Anthropic, and Cohere. For a limited time, tickets are just $292 for an entire day of expert talks, workshops, and potent networking.

### Exhibit at TechCrunch Sessions: AI

#### Secure your spot at TC Sessions: AI and show 1,200+ decision-makers what you’ve built — without the big spend. Available through May 9 or while tables last.

Berkeley, CA|June 5

[REGISTER NOW](https://techcrunch.com/events/tc-sessions-ai/?promo=tc_inline_exhibit&utm_campaign=tcsessionsai2025&utm_content=exhibit&utm_medium=ad&utm_source=tc)

OpenAI is increasingly relying on automated systems to mitigate the risks from its models. For example, to prevent [GPT-4o’s native image generator from creating child sexual abuse material (CSAM)](https://cdn.openai.com/11998be9-5319-4302-bfbf-1167e093f1fb/Native_Image_Generation_System_Card.pdf), OpenAI says it uses a reasoning monitor similar to the one the company deployed for o3 and o4-mini.

Yet several researchers have raised concerns OpenAI isn’t prioritizing safety as much as it should. One of the company’s red-teaming partners, Metr, said it had relatively little time to test o3 on a benchmark for deceptive behavior. Meanwhile, OpenAI decided not to release a [safety report for its GPT-4.1 model](https://techcrunch.com/2025/04/15/openai-ships-gpt-4-1-without-a-safety-report/), which launched earlier this week.

Topics

[AI](https://techcrunch.com/category/artificial-intelligence/), [ai safety](https://techcrunch.com/tag/ai-safety/), [ChatGPT](https://techcrunch.com/tag/chatgpt/), [OpenAI](https://techcrunch.com/tag/openai/)

![Image 6: Maxwell Zeff](https://techcrunch.com/wp-content/uploads/2024/07/Maxwell-Zeff.jpg?w=150)

Maxwell Zeff

Senior Reporter, Consumer

[](https://x.com/ZeffMax)

Maxwell Zeff is a senior reporter at TechCrunch specializing in AI and emerging technologies. Previously with Gizmodo, Bloomberg, and MSNBC, Zeff has covered the rise of AI and the Silicon Valley Bank crisis. He is based in San Francisco. When not reporting, he can be found hiking, biking, and exploring the Bay Area’s food scene.

[View Bio](https://techcrunch.com/author/maxwell-zeff/)

![Image 7: Event Logo](https://techcrunch.com/wp-content/uploads/2025/05/techcrunch-sessions-ai-small-logo.jpg)

June 5, 2025

Berkeley, California

LIMITED TIME: Save $300+ on your ticket, plus an additional 50% on a second for a full immersive day of AI! Hear from AI pioneers from Google DeepMind, OpenAI, Anthropic, and more on the main stage and in breakouts, and network like never before.

[REGISTER NOW](https://techcrunch.com/events/tc-sessions-ai/?utm_source=tc&utm_medium=ad&utm_campaign=tcsessionsai2025&utm_content=earlybirdbogo&promo=tceventssidebar_ebbogo&display=)

Most Popular
————

*
### [Apple could launch AI-powered smart glasses in 2026](https://techcrunch.com/2025/05/23/apple-could-launch-ai-powered-smart-glasses-in-2026/)

* [Dominic-Madori Davis](https://techcrunch.com/author/dominic-madori-davis/)

*
### [Trump threatens 25% tariffs on iPhones made outside the US](https://techcrunch.com/2025/05/23/trump-threatens-25-tariffs-on-iphones-made-outside-the-us/)

* [Dominic-Madori Davis](https://techcrunch.com/author/dominic-madori-davis/)

*
### [Mysterious hacking group Careto was run by the Spanish government, sources say](https://techcrunch.com/2025/05/23/mysterious-hacking-group-careto-was-run-by-the-spanish-government-sources-say/)

* [Lorenzo Franceschi-Bicchierai](https://techcrunch.com/author/lorenzo-franceschi-bicchierai/)

*
### [After Klarna, Zoom’s CEO also uses an AI avatar on quarterly call](https://techcrunch.com/2025/05/22/after-klarna-zooms-ceo-also-uses-an-ai-avatar-on-quarterly-call/)

* [Ivan Mehta](https://techcrunch.com/author/ivan-mehta/)

*
### [Anthropic CEO claims AI models hallucinate less than humans](https://techcrunch.com/2025/05/22/anthropic-ceo-claims-ai-models-hallucinate-less-than-humans/)

* [Maxwell Zeff](https://techcrunch.com/author/maxwell-zeff/)

*
### [Hinge Health pops 17%, but joins growing ranks of down-round IPOs](https://techcrunch.com/2025/05/22/hinge-health-pops-17-but-joins-growing-ranks-of-down-round-ipos/)

* [Marina Temkin](https://techcrunch.com/author/marina-temkin/)

*
### [Klarna CEO and Sutter Hill take victory lap after Jony Ive’s OpenAI deal](https://techcrunch.com/2025/05/22/klarna-ceo-and-sutter-hill-take-victory-lap-after-jony-ives-openai-deal/)

* [Julie Bort](https://techcrunch.com/author/julie-bort/)

Loading the next article

Error loading the next article

[![Image 8: TechCrunch Logo](https://techcrunch.com/wp-content/themes/tc-24/dist/svg/tc-logo.svg)](https://techcrunch.com/)
* [X](https://twitter.com/techcrunch)
* [LinkedIn](https://www.linkedin.com/company/techcrunch)
* [Facebook](https://www.facebook.com/techcrunch)
* [Instagram](https://instagram.com/techcrunch)
* [youTube](https://www.youtube.com/user/techcrunch)
* [Mastodon](https://mstdn.social/@TechCrunch)
* [Threads](https://www.threads.net/@techcrunch)
* [Bluesky](https://bsky.app/profile/techcrunch.com)

* [TechCrunch](https://techcrunch.com/)
* [Staff](https://techcrunch.com/about-techcrunch/)
* [Contact Us](https://techcrunch.com/contact-us/)
* [Advertise](https://techcrunch.com/advertise/)
* [Crunchboard Jobs](https://www.crunchboard.com/)
* [Site Map](https://techcrunch.com/site-map/)

* [Terms of Service](https://techcrunch.com/terms-of-service/)
* [Privacy Policy](https://techcrunch.com/privacy-policy/)
* [RSS Terms of Use](https://techcrunch.com/rss-terms-of-use/)
* [Code of Conduct](https://techcrunch.com/code-of-conduct/)

* [Anthropic](https://techcrunch.com/2025/04/30/anthropic-suggests-tweaks-to-proposed-u-s-ai-chip-export-controls/)
* [OpenAI](https://techcrunch.com/2025/04/29/openai-explains-why-chatgpt-became-too-sycophantic/)
* [Duolingo](https://techcrunch.com/2025/04/30/duolingo-launches-148-courses-created-with-ai-after-sharing-plans-to-replace-contractors-with-ai/)
* [Wikipedia](https://techcrunch.com/2025/04/30/__trashed-4/)
* [Hugging Face](https://techcrunch.com/2025/04/28/hugging-face-releases-a-3d-printed-robotic-arm-starting-at-100/)
* [Tech Layoffs](https://techcrunch.com/2025/02/28/tech-layoffs-2024-list/)
* [ChatGPT](https://techcrunch.com/2025/01/28/chatgpt-everything-to-know-about-the-ai-chatbot/)

![Image 9](https://pixel.wp.com/g.gif?v=ext&blog=136296444&post=2995065&tz=-7&srv=techcrunch.com&hp=vip&j=1%3A14.6&host=techcrunch.com&ref=&fcp=392&rand=0.2706418549648887)

Some areas of this page may shift around if you resize the browser window. Be sure to check heading and document order.

AI Track Hub

AI Track Hub

OpenAI’s latest AI models have a new safeguard to prevent biorisks

You May Also Like

OpenAI’s startup empire: The companies backed by its venture fund

OpenAI debuts Codex CLI, an open source coding tool for terminals

ChatGPT Search is not OpenAI’s ‘Google killer’ yet