AI news Oct 30: AWS Growth Accelerates as Attacks Hit AI Developer Tools

Amazon's cloud unit posted its fastest growth in nearly five years, and enterprise AI vendors kept reporting strong demand. At the same time, a run of incidents targeting AI developer tools put security in focus.
- AWS grew 39.3% in Q3 with a $578 billion backlog, and Amazon's chips business is growing at triple digits.
- Wiz says at least nine AI developer platforms reported breaches in October, with stolen keys being resold.
- A fake Claude Code extension stayed in Cursor's marketplace for 11 hours.
- Toyota and Waymo plan autonomous bZ4X tests in Japan from 2027.
- SAP introduced AI governance tools, including the Enterprise AI Gateway.
Cloud and enterprise AI demand
AWS grew 39.3% year over year in the third quarter, its sixth straight quarter of acceleration, according to Andy Jassy. He said backlog stands at $578 billion. Jassy also said Amazon's chips business is growing at triple-digit percentages, with revenue commitments for Graviton and Trainium each up more than 20x from a year earlier.
Sierra says it has passed $400M in ARR, adding over $150M in the third quarter alone. At SAP TechEd, SAP introduced AI governance tools led by the Enterprise AI Gateway, including an assistant for EU AI Act assessments. Databricks and SAP also expanded their partnership to bring Genie and Lakebase to SAP data. Separately, Cognizant formed a group of more than 25,000 professionals focused on delivering OpenAI products.
Security incidents around AI tooling
Wiz says at least nine AI developer platforms disclosed breaches in October, and that stolen credentials are already being resold. In two enterprise cases the stolen keys drove roughly 2.5 million model calls.
A tampered build of Anthropic's Claude Code extension, version 2.1.313, sat in Cursor's marketplace for 11 hours on October 29 and tried to install an SSH backdoor. Up to 5,000 users may have installed it. Meanwhile, a Brotli decoder flaw nicknamed Node2Shell (CVE-2026-111497) can lead to remote code execution and has been added to CISA's Known Exploited Vulnerabilities catalog.
On the defence side, CrowdStrike detailed Agent Graph for tracing AI agent actions and extended OverWatch threat hunting to AI agents and Google Cloud.
Autonomous vehicles and science
Toyota and Waymo will start testing an autonomous bZ4X fleet in Japan in 2027, with the United Kingdom and other Waymo markets planned afterwards. Kodiak AI says its driverless readiness measure has reached 99%, with a first route on Interstate 45 between Dallas and Houston. EPFL researchers argue that robotaxis depend more on regulation, acceptance and public transport integration than on technology.
In research, Zhejiang Lab presented its 021 Scientific Foundation Model for chemistry, genomics, astronomy and other fields. Breakthrough Listen and the SETI Institute are using AI agents to steer the Allen Telescope Array. Envision's Tianji model ranked first in ECMWF's AI Weather Quest, and Epoch AI found that 17% of biology preprints acknowledge LLM use.
Policy, safety and developer tools
OpenAI cited amicus briefs in the New York Times case to defend training as fair use. It also said it will honor a shared robots.txt signal and added violence safeguards to ChatGPT. Singapore's Chief Justice said human judgment must remain central in courts, the ITU warned of an AI divide, and Resemble AI's report tracked deepfakes in election news.
In tooling, NVIDIA detailed NemoClaw, Factory acquired Typedef, Hidden Door launched an MCP server, Mintlify described its self-updating docs design, and VAST released Event Broker 5.5.

