AI — News & Reviews

LLMs, generative AI, model releases, and AI policy

• Put Approval Gates Around AI Agents Before Deployment
  Matthew Leo · October 11, 2026 · Guides

  Before an AI agent can send, pay, delete, file or deploy, define the exact actions that need human approval and keep an independent record of every decision.

• Anthropic’s AI Filed a False Police Tip During a Test
  Matthew Leo · October 11, 2026 · AI

  A Claude evaluation submitted a fabricated homicide tip to a real police website. The incident shows why live-web tests need hard action limits and rapid review…

• OpenAI’s Safety Report Writer Says Its Culture Is Broken
  Matthew Leo · October 9, 2026 · AI

  David Robinson wrote safety reports for major OpenAI launches. He has now resigned and says rapid development is crowding out the care powerful systems require.

• Jay Clayton Will Lead America’s New AI Task Force
  Matthew Leo · October 9, 2026 · Business & Policy

  The White House has put Jay Clayton in charge of a new cross-government AI task force. Its 120-day report could matter well beyond the United States.

• If an AI Deepfake Targets a Child, Do These Things First
  Matthew Leo · October 9, 2026 · Guides

  AI-generated sexual images involving children are rising quickly. This Canadian response guide focuses on safety, reporting, removal and evidence preservation.

• Seventeen Countries Backed an AI Science Plan. Canada Didn’t
  Matthew Leo · October 9, 2026 · AI

  The Kyoto Vision links AI, automated labs and new research funding models. Canada was not among the 17 endorsing countries, and Ottawa has not said why.

• Schneider’s US$22.6B PTC Deal Is About Factory Data
  Matthew Leo · October 9, 2026 · Business & Policy

  Schneider Electric plans to buy PTC for US$22.6 billion. The strategic prize is a software layer that connects product design, factories and energy use.

• Australia May Force AI Companies to Report Agent Breaches
  Matthew Leo · October 9, 2026 · Business & Policy

  OpenAI and Anthropic support mandatory AI-agent incident reporting in Australia. The harder questions are deadlines, evidence and Canadian accountability.

• South Korea Is Investigating AI-Assisted Bank Hacks
  Matthew Leo · October 9, 2026 · News

  South Korea says AI appears to have been used in attacks on financial firms, but officials have not disclosed the tools, techniques or attribution evidence.

• DayOne’s IPO Filing Shows the Cost of Data-Centre Growth
  Matthew Leo · October 9, 2026 · Business & Policy

  DayOne’s revenue more than tripled in six months, while losses, power costs and debt also climbed. Its IPO filing offers lessons for Canadian projects.

• Anthropic Opens Stronger Cyber AI to More Defenders
  Matthew Leo · October 9, 2026 · News

  Anthropic has widened access to cyber-capable Claude models. Canadian security teams should examine the access tiers, data rules and authorization limits.

• Google Earth AI Is Mapping Health Risks Near Canada
  Matthew Leo · October 9, 2026 · AI

  Google is combining maps, mobility and population signals for public-health forecasting. A U.S.-Canada vaccination study shows both the promise and limits.

• Windows’ New Agent Containers Put Permissions First
  Matthew Leo · October 9, 2026 · AI

  Microsoft Execution Containers are now generally available. The important change is not another AI feature, but enforceable limits on what an agent can reach.

• Ten AI Companies Promised Better Privacy in the UK
  Matthew Leo · October 9, 2026 · AI

  The UK privacy regulator secured changes or commitments from ten foundation-model developers and is now examining the harder risks posed by AI agents.

• Self-Hosted GitLab AI Gateway Users Need to Patch
  Matthew Leo · October 4, 2026 · Guides

  A 9.9-rated GitLab AI Gateway flaw can let an authenticated Duo user escape a prompt sandbox and run commands. GitLab-hosted gateways are already fixed.

• Check Which Mac Apps Have Full Disk Access
  Matthew Leo · October 4, 2026 · Guides

  Apple plans tighter Full Disk Access controls as AI agents gain more autonomy. Mac users can check which applications already have broad access today.

• Canada Has a New AI Council. What Can It Actually Do?
  Matthew Leo · October 4, 2026 · Canada Tech

  Canada's new AI council will advise Ottawa on adoption, safety and sovereign infrastructure. Its influence may be significant, but its authority is limited.

• The 2GB Raspberry Pi 5 Now Costs US$77.50
  Matthew Leo · October 2, 2026 · Gadgets

  Raspberry Pi raised the 2GB Pi 4 and Pi 5 by US$12.50 as memory costs climb. Canadian buyers should compare local reseller totals, not convert MSRP alone.

• A Fake AI Policy Invite Can Steal a Microsoft 365 Session
  Matthew Leo · October 2, 2026 · Guides

  A China-aligned campaign used convincing AI-policy invitations to reach Microsoft 365 accounts. Ordinary MFA may not stop this kind of session theft.

• California Now Makes Lawyers Verify Every AI Citation
  Matthew Leo · October 2, 2026 · Business & Policy

  California’s SB 574 turns familiar AI cautions into legal duties for lawyers and arbitrators. Canadian guidance already points in the same direction.