AI Security

Anthropic outlines customer-controlled monitoring option for Claude

Anthropic Introduces Enterprise Frontier Safeguards for Claude Image: Primary
Anthropic said it introduced Enterprise Frontier Safeguards on September 1, an option intended to pair zero data retention with automated monitoring for serious misuse in Claude deployments. The company said eligible customers could hold related monitoring data in their own Amazon S3, Azure Blob Storage or Google Cloud Storage accounts, using their own encryption keys, access policies and audit logs. Anthropic said automated systems would analyze a rolling traffic window and send signals to customer personnel for review, without Anthropic employee review. Phased access is planned later this fall.
Sources
letsdatascience.com
In this story
Anthropic Google Microsoft Amazon
Published by Tech & Business, a media brand covering technology and business. This story was sourced from letsdatascience.com and reviewed by the T&B editorial agent team.
Back to Newswire
Keep reading
Full wire
Security
Security

SonicWall reports exploited SMA1000 zero-days and releases fixes

SecurityWeek reported that SonicWall urged SMA1000 customers to patch two zero-day vulnerabilities the vendor says have been exploited. CVE-2026-83548 is a pre-authentication SSRF flaw in the Appliance Work Place interface that ca...

SecurityWeek
Infrastructure Products
Infrastructure Products

HyImpulse raises more than €50 million in Series A extension

HyImpulse Technologies announced a Series A extension of more than €50 million in equity for its suborbital and orbital launch-services business. The round was co-led by JOIN Capital and Ace Capital Partners, with new and existin...

EU-Startups
Science AI
Science AI

Preprint tests text-isolation method for multimodal model errors

A preprint describes a 998-case diagnostic for cases in which external text overrides conflicting image evidence in multimodal language models. In its tests, the proposed System-2 Visual Arbitration method withheld text from a vi...

cs.AI updates on arXiv.org
AI Science
AI Science

Preprint reports iterative framework lifts coding-agent benchmark results

A preprint describes Harness-of-Harness, a framework that organizes coding-agent harnesses into iterative planning, coding and testing loops intended to improve autonomous software development. The authors report average relative ...

cs.AI updates on arXiv.org
Security Policy
Security Policy

Authorities disrupt Sality botnet infrastructure in joint action

International law-enforcement agencies and private partners seized Sality-linked domains and used a peer-to-peer sinkhole operation to disrupt the Sality botnet, according to BleepingComputer. The operation isolated infected machi...

BleepingComputer
AI
AI

OpenAI says ChatGPT Health is integrating with Epic

OpenAI said it integrated ChatGPT Health with Epic's electronic health record system and introduced a Healthcare Public Data plug-in that can retrieve information from sources including PubMed. The supplied summary does not specif...

Techmeme