7 hrs ago
Anthropic Changes Enterprise Data Retention Rules With New Safeguards
Anthropic is changing how it handles data from business customers.
It introduced a 30-day storage rule for traffic involving its advanced Mythos-class models in June.
The rule was intended to help find misuse and defend against attacks.
Businesses were concerned about privacy and controlling their data.
Anthropic has created a system called Enterprise Frontier Safeguards.
It lets companies decide more directly how their data is reviewed and stored.
Safety checks can happen automatically on a company’s own systems.
The system will be introduced in stages and is expected to be generally available in fall 2026.
Anthropic announced Enterprise Frontier Safeguards after feedback from enterprise customers.
The system will let businesses manage data review, storage and oversight.
Automated safety checks can run without review by Anthropic staff.
Customers can monitor data on their own systems without giving Anthropic access to underlying data.
The safeguards will launch in phases, with general availability planned for fall 2026.
- Who
- Anthropic and its enterprise customers.
- What
- Anthropic announced Enterprise Frontier Safeguards, a data-management and automated safety-monitoring system for business customers.
- Where
- Monitoring can take place on customers’ own systems, and the safeguards can support direct or cloud-provider deployments.
- When
- The announcement was made Tuesday; general availability is planned for fall 2026.
- Why
- Anthropic said the change responds to enterprise feedback about privacy and data control while preserving safety testing and monitoring.
Enterprise Data Control
Safety Monitoring And Retention
Data retention
Enterprise Data Control
Enterprise customers sought greater control over how their data is reviewed and stored, according to Anthropic’s announcement.
Safety Monitoring And Retention
Anthropic introduced the 30-day retention requirement to help identify misuse and defend against advanced and evolving attacks.
Access to underlying data
Enterprise Data Control
Companies can conduct safety testing on their own systems without giving Anthropic access to the underlying data.
Safety Monitoring And Retention
Anthropic’s earlier policy kept information for safety purposes, although the company said it would not use it for other purposes such as training its AI models.
Key facts
- New solution
- Enterprise Frontier Safeguards
- Affected users
- Enterprise customers using Anthropic’s advanced models
- Previous policy
- A 30-day data-retention requirement introduced in June for traffic involving Mythos-class models
- Data access
- Businesses can perform mandatory safety testing without giving Anthropic access to the underlying data
- Human review
- Automated safety checks will run without requiring review by Anthropic staff
- Customer charge
- Anthropic said it will not charge customers for Enterprise Frontier Safeguards
- Availability
- Phased launch, with general availability planned for fall 2026
Sources
Related news
OpenAI Tests Zero-Retention Safety as Anthropic Retains Data
Anthropic Releases Fable 5.1 and Mythos 5.1
OpenAI Tests Private Safety Processing For Paid Customers
OpenAI Previews Privacy-Preserving System to Monitor AI Misuse
Claude Cowork and Chat Merge Memory for Smoother Workflows
Anthropic launches Claude Fable and Mythos 5.1 models
OpenAI’s Astra Reaches Critical Cybersecurity Threshold Ahead of Limited Rollout
Google Prepares Gemini 3.8 Flash to Challenge AI Rivals
