AI NEWSROOMBY CLASSWISE.IO

Monday, August 10, 2026

Claude Code sets auto mode default

Plus: Adults use AI for finance despite low trust and researchers design functional viruses.

DEVELOPER TOOLS

🤖 Claude Code defaults to auto mode

Claude Code auto mode becomes default for Pro, Max, and Team users on August 14. Auto mode uses a classifier to check tool calls for irreversible or destructive actions.

The details:

  • Anthropic will no longer charge extra tokens for the auto mode classifier usage.
  • A study showed auto mode caught 89% of dangerous commands versus 13.6% for humans.
  • Team and Enterprise customers using auto mode ship about 25% more pull requests.

Why it matters: The demonstrated capability is bounded by its tested scope. Auto mode caught dangerous commands at a high rate in a study, and it falls back to manual approvals after repeated blocks.

First reported by 9to5mac.com, 57h ago · 9to5mac.com · techcrunch.com · techinasia.com · indiatoday.in

FINANCE

📉 Adults use AI for finance despite low trust

Adults use AI for finance despite low trust
Image source: ktvn.com

Gallup found that only three percent of U.S. adults trust AI a great deal. The poll found that about a quarter of Gen Z and millennial adults who looked for financial advice used AI.

The details:

  • Gallup conducted the survey in partnership with financial services firm Edward Jones.
  • About 73% of U.S. adults who sought financial advice relied on their own internet research.
  • About 8 in 10 U.S. adults have at least some confidence in financial advisers.

Why it matters: Reported use is not reported trust. Younger adults who sought financial advice used AI, while Gallup found that few U.S. adults trust the technology a great deal.

First reported by ktvn.com, 64h ago · ktvn.com · wral.com · scrippsnews.com · news.google.com

BIOTECH

🧬 AI designs functional viruses

AI designs functional viruses
Image source: bbc.com

US researchers used artificial intelligence to design fully functional replicating viruses. The AI models were trained on genetic codes from viruses, bacteria, plants and people.

The details:

  • This marks the first time AI successfully designed whole genomes.
  • Sixteen novel viruses were created to infect bacteria without threatening humans.
  • Stanford researchers synthesised 302 promising AI-generated virus designs in the laboratory.

Why it matters: The demonstrated capability is bounded by its tested scope. AI designed whole genomes, but the training database excluded viruses that could infect complex organisms and the resulting viruses targeted only bacteria.

First reported by bbc.com, 82h ago · bbc.com · 2oceansvibe.com · wionews.com

📰 Everything else in AI today

🛠️ Tool of the day

Snowflake Snowflake integrates with Saviynt to manage AI agent security risks within enterprise data platforms.

Researched and written by AI Newsroom, by Classwise.io

See how AI Newsroom is researched and checked

Get tomorrow's issue in your inbox.

Free, daily, 4:57 AM ET. Five minutes and you know what matters in AI.

Join for free