OpenAI has begun previewing a new privacy-focused safety service called Private Safety Processing, a move that escalates its competitive rivalry with Anthropic over enterprise customer trust. The system, now being tested with select customers, is designed to monitor for AI misuse across multiple conversations without retaining any of the customer’s data — a direct counter to Anthropic’s recently implemented 30-day data retention policy for its most advanced models.
The announcement comes as both AI labs aggressively court enterprise clients, a market where data privacy and security are decisive factors. OpenAI’s approach leverages its existing Zero Data Retention (ZDR) framework, which already allows API customers to have their data discarded after each session while still enabling automated abuse monitoring.
Also read: SpaceX's Reported Bid for AI Coding Startup Cognition Draws Denial from CEO
How Private Safety Processing works
Private Safety Processing extends ZDR’s scope by enabling what OpenAI describes as “long-horizon safety monitoring.” Instead of examining individual sessions in isolation, the system assesses inputs and outputs across multiple conversations, allowing it to detect patterns of malicious use that would be invisible in a single session.
According to a TechCrunch report, the technology is particularly aimed at bad actors who deliberately spread their requests across multiple sessions to evade detection — for instance, someone incrementally engineering malware or probing for vulnerabilities. The monitoring is conducted by automated agents, which, if triggered, can analyze cross-session activity for signs of abuse without human review of the user’s conversations.
When the system detects a potential issue, it sends a “narrowly defined signal” to OpenAI indicating a specific type of suspicious activity. The company can then decide whether enforcement is necessary. If so, OpenAI may reach out to the customer for additional context, and the customer can choose to share data at their discretion.
Anthropic’s contrasting approach
Anthropic’s July 2026 data retention policy, which applies to all “covered models” including its Mythos-class systems and future models with similar capabilities, has drawn criticism from enterprises that handle sensitive information. The policy allows Anthropic to retain all user sessions and conversations for 30 days to enable safety analysis.
While Anthropic maintains that human review of customer data can only occur through a “controlled access path” involving a small set of approved reviewers — with every session recorded in a tamper-proof log — the retention itself has been a sticking point for businesses with strict data governance requirements.
Both companies largely adhere to ZDR principles, but Anthropic’s exception for covered models has created a clear differentiation in the market. OpenAI’s Private Safety Processing appears designed to capitalize on that distinction, positioning itself as the privacy-preserving alternative for enterprises that cannot afford to have their AI interactions stored.
What this means for enterprise AI adoption
The timing is significant. A recent report indicated that OpenAI’s Q2 growth lagged behind Anthropic’s, with Anthropic’s annualized revenue run rate reportedly reaching $65 billion. Anthropic investors have floated a potential IPO at a $2 trillion valuation, while OpenAI is also reportedly working on its own public offering.
For enterprises, the practical implication is a widening array of choices in how AI vendors handle data privacy. OpenAI’s approach offers the promise of comprehensive safety monitoring without data retention, which could be particularly appealing to regulated industries such as healthcare, finance, and legal services.
However, questions remain about the effectiveness of automated cross-session monitoring compared to human review. OpenAI’s system, while privacy-preserving, may be less flexible in nuanced investigations. The company says it will continue to refine the technology based on feedback from its preview customers.
As the AI arms race intensifies, the battle over data privacy policies is likely to become a central battleground. Both OpenAI and Anthropic are betting that their respective approaches will win over enterprise clients — and the market’s response could shape industry standards for years to come.
This article discusses corporate strategies and product announcements. It does not constitute financial advice, and the AI market remains volatile and uncertain. Readers should conduct their own research before making investment decisions.
Be the first to comment