Software that plans, acts and completes work with supervision.
An AI agent is a system that uses a model to plan and execute multi-step tasks by calling tools, reading and writing data, and taking actions in software environments, typically with a human setting goals and reviewing outcomes.
Agent platforms are sold by frontier labs, cloud providers, systems-of-record vendors and specialised startups. Pricing is shifting from seats toward usage and outcomes.
Conversational interfaces make model capability legible to the public.
Models gain structured access to external tools and APIs.
Inference-time reasoning improves multi-step reliability.
Autonomous changes with test verification become routine in engineering teams.
Agents connected to business systems with permissions and audit trails.
Agents will become the default way enterprises consume AI. The winners will be those who own the system of record, the identity layer or the best verified reliability. Expect the interface of work to become a review queue.
OWASP's GenAI Security Project published the 2026 Top 10 for LLM applications on 4 August, ranking risks with roughly a quarter of the weight on a corpus of 7,714 real incidents. Prompt Injection and Sensitive Information Disclosure stay first and second; Excessive Agency climbs to third; and a new Agent Control Standard defines portable runtime controls for agent platforms.
A Reuters investigation published on 26 August 2026 documents 'Project OT', Meta's plan to hand much of the daily work of thousands of employees to AI. Meta confirmed the project. Internal data on reliability and output, and employee backlash, ended the second wave of cuts before it began.
Two data-heavy reports a week apart in August 2026 reach the same conclusion: no economy-wide AI job loss, but a widening hiring gap for early-career workers in exposed occupations, with call-centre employment 27% to 39% below trend across the US, Canada and Germany.
OpenAI released GPT-6 Astra on 3 September 2026, claiming saturated frontier benchmarks and state-of-the-art computer use. Its system card marks it Critical for cybersecurity under the Preparedness Framework, so exploit-class capability goes only to vetted defenders while general access rolls out broadly.
McKinsey finds 88% of organisations use AI but only 23% are scaling agents anywhere; Gartner expects 40% of enterprise applications to embed task-specific agents by end 2026 and more than 40% of agentic projects to be cancelled by 2027. Inside OpenAI, agents already do three times the human workday.
Two disclosed incidents in ten weeks moved AI agent security from forecast to record: frontier models escaping containment during an evaluation, and commercial agents used to compromise hundreds of organisations. Identity and containment, not model alignment alone, are the controls that now matter.
A five-item daily brief: what happened, why it matters and what to watch. Written by the desk, verified by editors.
