AI Weekly Malaysia

OpenAI announced Private Safety Processing, a mechanism that automatically scans customer model interactions for safety risks while maintaining Zero Data Retention (ZDR) commitments—something rival Anthropic hasn't achieved for its top models (Mythos 5 and Fable 5), which require 30-day prompt and output retention as of June 9, 2026. OpenAI says automated systems return limited safety signals without exposing underlying prompts or responses to its personnel, and will soon offer customer-controlled encryption keys, though technical details remain unpublished.

Why: If you're building on Anthropic's top-tier models under a ZDR agreement, your prompts and outputs are actually retained for 30 days—this may violate your own enterprise data contracts. Teams evaluating OpenAI vs Anthropic for privacy-sensitive workloads should factor this gap into procurement decisions, but should wait for OpenAI to publish Private Safety Processing technical details before assuming true ZDR, since the mechanism is currently undescribed.

OpenAI is previewing 'Private Safety Processing' to select customers, an automated abuse-monitoring system that retains zero customer data, positioned as a counter to Anthropic's July-announced data-retention policy. Anthropic's policy retains all user sessions for 30 days for 'covered models' (Mythos-class and similar, including Fable), which has frustrated enterprise customers handling sensitive data. OpenAI already follows Zero Data Retention (ZDR) via per-session API agents, and says this new technology widens ZDR's scope.

Why: If you're building on Anthropic's covered models (Mythos-class, Fable) for enterprise clients with data sensitivity or compliance constraints, factor in the 30-day session retention when choosing your provider. OpenAI's Private Safety Processing preview gives a concrete alternative if zero retention is a hard requirement — but it's still in preview for select customers, so verify availability before committing.

Ramp launched Router, an AI model routing API similar to OpenRouter, offering access to models from OpenAI, Anthropic, DeepSeek, Moonshot, Minimax, Nvidia, xAI, and Z.ai with routing strategies based on cost, benchmarks, or difficulty. It is US-only, free for the rest of 2026 (excluding inference costs) with a $26 launch credit, and has a default-on data retention policy that logs inputs, outputs, and tool calls for one year unless users opt out.

Why: Router is US-only, so Malaysian builders cannot use it directly, but its default-on one-year data retention policy is a concrete reminder to check what your model router logs by default—especially if you route sensitive customer data through third-party routers. The free-through-2026 pricing and $26 credit are temporary and pricing for 2027 is undisclosed, so don't build long-term cost assumptions on it.

Flock, a US surveillance company operating automated license plate readers (ALPRs), faces growing backlash after a Washington Post investigation documented 46 cases of police officers abusing the system, including stalking former partners. CEO Garrett Langley apologized and reduced default data retention from 30 days to 7 days, though a new 'Evidence Mode' lets law enforcement keep data longer. An online 'De-Flock America' campaign is encouraging people to vandalize Flock cameras on Halloween 2026.

Why: This is a US law-enforcement privacy story with no direct impact on builders in this audience. The only tangentially relevant detail is the data retention reduction (30→7 days) and the tension between default privacy settings and customer overrides—a pattern relevant to anyone designing systems that store sensitive location data, but the article lacks technical depth to act on.