Goldman Sachs reins in AI coding tools, Argentina courts AI firms with no rules
Goldman Sachs pulls AI coding tools from non-engineers just months after praising the practice, Argentina pitches itself as a regulation-free zone for AI firms, and the New York Times sues browser maker Brave over scraped content.
Every line links to a primary source in the full briefing.
01Goldman Sachs walks back open AI coding access for non-engineers
Goldman Sachs has restricted access to AI coding tools such as Claude Code and GitHub Copilot to engineers only, after months of letting traders and bankers use them to build prototypes and monitor trading flows. The bank cites productivity and risk management, a reversal from July when its CIO publicly praised non-technical staff for 'vibe coding'.
A major bank deciding that letting non-technical staff write and ship code unsupervised was too risky is a useful data point for any company that has handed AI coding tools to people outside its engineering team.
Source: The Decoder
02Argentina pitches itself as a no-rules zone for AI companies
Argentina's president is positioning the country as a haven for AI firms by promising an absence of regulation, part of a broader push to attract AI investment and infrastructure.
Watch for similar jurisdictions competing on looser rules rather than incentives, which could shift where AI infrastructure and data processing happens, and what legal protections apply to the data involved.
Source: Broadband Breakfast
03New York Times sues Brave over AI content scraping
The New York Times, along with the Daily News, Chicago Tribune and Orlando Sentinel, sued browser maker Brave, accusing it of scraping copyrighted articles with an undisclosed crawler and distributing them to AI companies including Anthropic, Perplexity, Mistral, xAI and Cohere. The suit also targets a Brave feature that bypasses paywalls.
Any business relying on AI tools built on scraped content should expect more legal fights like this one, and possibly disruption or new licensing costs passed down the chain.
Source: TechCrunch
04OpenAI boosts bug bounty payouts and runs an internal security sprint
OpenAI is increasing its security bug bounty rewards and adding bonuses for report quality and loyal researchers, while separately running a new internal exercise with Cathedral Systems to test its own defenses using AI.
It is a reminder that even well-resourced AI vendors treat their own security as unfinished work, so businesses should not assume a provider's infrastructure is fully hardened by default.
05Commentary: regulators should block Anthropic's IPO
An opinion piece argues the SEC should block Anthropic's planned IPO on public protection grounds, adding to a week of scrutiny over the company's valuation swings and its own IPO filing warning about risks from the AI it builds.
Growing public debate over how AI companies are valued and governed ahead of going public is worth tracking for any business that depends on one of these vendors staying stable and well capitalised.
Source: Tech Policy Press
Review who at your company can use AI coding tools unsupervised
Goldman Sachs just reversed course on letting non-engineers use AI coding tools after finding it created productivity and risk problems. Take an hour this week to check which of your staff have access to AI coding assistants and whether their output gets reviewed before it touches anything live.
Sources
6 links- 01Argentina’s Milei Pitches a Regulation-Free AI Haven at the End of the WorldBroadband Breakfast
- 02Rewarding defenders with bigger bug bountiesOpenAI
- 03Preparing our next security sprint to harden OpenAIOpenAI
- 04New York Times and other outlets sue web browser Brave over alleged content scrapingTechCrunch
- 05The SEC Should Block Anthropic’s IPO to Protect the PublicTech Policy Press
- 06Goldman Sachs pulls AI coding tools from traders and bankersThe Decoder