AI moves code review into the editor, reducing routine human checks

Small engineering teams will spend less time on trivial code mistakes but must add audit and governance work; operations face faster agent sprawl, regulatory uncertainty and new security overhead.

Claude shifts code review from a gate to a continuous pass

Researchers on the Claude Code team say the question has moved from “did a human read the code?” to which check should catch each class of problem. Claude reviews code while engineers write it, catching routine mistakes so many changes are already fixed by the model before a human review happens.

Why it matters: Teams will reclaim reviewer hours previously spent on trivial fixes but must build logging and step‑in rules for anything the model changes or suggests, because human attention will concentrate on higher‑level decisions.

Stack Overflow for Agents ships fast and growth spikes with a ChatGPT plugin

OpenAI reports two engineers built Stack Overflow for Agents in a few weeks using Codex, and its ChatGPT plugin helped triple signups for the service while driving up to 10× growth in agent creation. The project shows how quickly teams can stand up agent tooling and get users to build agents at scale.

Why it matters: Rapid agent creation reduces time to experiment but creates operational risk from agent sprawl, hidden costs and unexpected behaviour; teams need inventory, rate limits and approval workflows for new agents.

Itaú deploys generative AI at scale with a governance platform

Itaú is rolling out ia.i, a conversational interface, alongside a governance platform called IARA to apply its AI First strategy across customers and employees in a tightly regulated market. The bank pairs customer‑facing generative features with internal governance controls rather than leaving deployment ad hoc.

Why it matters: Operators selling to financial services or regulated industries should expect buyers to demand built‑in governance and auditability rather than one‑off models; implementing governance early reduces rework and compliance risk.

Schools and libraries push back as FCC considers ending E‑Rate discounts

School districts and libraries urged the FCC to abandon a proposal that would limit or sunset E‑Rate, the $2 billion‑a‑year programme that subsidises Internet access for students and visitors. The FCC issued a Notice of Proposed Rulemaking that could cut discounts and restrict how funds are used, and the chair signalled the programme may be narrowed or ended.

Why it matters: Organisations that supply connectivity, hardware or services to public education or libraries face higher prices or tighter procurement rules if E‑Rate is reduced; factor increased telecom costs into bids and contracts if your customers rely on the programme.

FDA review may give 'enforcement discretion' for some peptide products

A filing shows the FDA sent an