NYU’s Gary Marcus on Building AI We Can Trust
THE SO WHAT
Trustworthy AI is moving from academic panel topic to boardroom KPI—Marcus is giving language to concerns your risk and compliance teams already have. If you ship AI features, expect more pointed questions on failure modes, auditability, and who is accountable when the system is wrong.
READ THE SOURCE
MORE FROM THE WIRE
Applied AIHostinger Reach email hosting review
AI shortcuts in email tools are now good enough for drafting, not yet for fully autonomous lifecycle management. If your brand or revenue depends on a newsletter, keep humans on strategy, segmentation, and final send while you quietly offload the grunt work.
Applied AINorrsken and Nvidia to open Stockholm AI Lab for founders in 2027
Norrsken and Nvidia planting an AI hub in central Stockholm extends the “compute plus ecosystem” model into the Nordics. If you’re a European founder in climate, health, or food, this is a signal to map your talent and capital strategy around emerging regional clusters, not just London and Berlin.
Applied AIChainalysis: agentic commerce on Coinbase's Base grew from ~1,500 transactions in Q2 2025 to 138M+ in Q3 2026; 87%+ of the payments were for less than 10 cents
138M+ mostly sub–10 cent transactions on Base in a year shows agentic commerce is real at machine scale, not a demo. If you sell digital goods or run APIs, start designing SKUs, pricing, and rate limits for autonomous agents making millions of micro-purchases, not humans clicking checkout.
Bank of America hired contractors to protect its software from powerful rogue AI agents
Bank of America hiring 1,000 people for “Mythos remediation” is a clear tell—large enterprises now treat rogue AI agents as an operational risk class on par with major cyber threats. If you’re deploying agents internally, budget for hardening and red-teaming as a first-order cost, not an afterthought.