Home / Security updates / Read our latest LLM safety and security publication: "Detecting misbehavior in frontier reasoning models"
Read our latest LLM safety and security publication: "Detecting misbehavior in frontier reasoning models"
Security updates detail rendered from /security-updates/upd_755edad8a91850d4.
Overview
| ID | upd_755edad8a91850d4 |
| Collection | Security Updates |
| Provider | SafeBase |
| Company | OpenAI |
| URL | - |
| Counts | - |
| Updated | - |
Raw record
| Field | Value |
|---|---|
| id | upd_755edad8a91850d4 |
| providerId | safebase |
| organizationId | org_c6356a1ea58bf030 |
| trustCenterId | tc_04d29d6479058309 |
| title | Read our latest LLM safety and security publication: "Detecting misbehavior in frontier reasoning models" |
| message | Frontier reasoning models exploit loopholes when given the chance. We show we can detect exploits using an LLM to monitor their chains-of-thought. Penalizing their “bad thoughts” doesn’t stop the majority of misbehavior—it makes them hide their intent. Read more on our [blog](https://openai.com/index/chain-of-thought-monitoring/)! |
| url | - |
| publishedAt | 2025-03-10 |
| source | {
"field": "statuspage/public/compliance-update",
"category": "general"
} |
| company | {
"id": "org_c6356a1ea58bf030",
"name": "OpenAI",
"domains": [
"trust.openai.com",
"openai.com"
]
} |
| trust_center | {
"id": "tc_04d29d6479058309",
"name": "OpenAI",
"url": "https://trust.openai.com",
"host": "trust.openai.com"
} |
| provider | {
"id": "safebase",
"name": "SafeBase"
} |
| links | {
"self": "/v1/security-updates/upd_755edad8a91850d4",
"company": "/v1/companies/org_c6356a1ea58bf030",
"trust_center": "/v1/trust-centers/tc_04d29d6479058309",
"provider": "/v1/providers/safebase"
} |
Get this page with API
Rendered from the bluedoor Trust Centers API. Reproduce it:
GET https://api.bluedoor.sh/trust-centers/v1/security-updates/upd_755edad8a91850d4JSON