1. Agentic AI Startup Repodo Automates Financial Auditing for Banks
FinTech Futures · Finance & Banking
Repodo, a fintech startup, has built an operating system for financial auditing that uses agentic AI to automate data collection, reconciliation, documentation, and transaction analysis — while leaving final sign-off to human judgement. The company was highlighted as one of the week’s top fintech stories on 28 August 2026, reflecting growing industry interest in AI that can handle complex, multi-step financial compliance tasks. The move signals a broader shift toward agentic AI taking on back-office banking work that has traditionally required significant manual effort.
2. California’s Attorney AI Bill Races to Final Vote Before Deadline
Artificial Lawyer · Law & Legal
California’s SB 574, a bill that would impose binding legal guardrails on how attorneys and arbitrators use generative AI, is in its final days as the legislature’s session closes August 31. The bill would require lawyers to personally verify AI-generated citations, protect client confidentiality from public AI systems, and prohibit arbitrators from delegating any decision-making to AI. It was spurred by a wave of AI hallucination incidents in courts nationwide, with legal researchers tracking more than 600 cases of attorneys submitting briefs containing fabricated case citations.
3. EU AI Act High-Risk Pharma Rules Now Live, Sector Scrambles to Comply
Drug Target Review · Healthcare & Pharmaceutical
The EU AI Act’s high-risk provisions came into full effect on 2 August 2026, placing strict compliance obligations on pharmaceutical and medical device companies using AI in clinical decisions, diagnostics, and drug development. The pharmaceutical sector is explicitly listed in Annex III of the Act, meaning AI tools that influence patient care or clinical outcomes must now meet rigorous transparency, documentation, and oversight standards. Companies that have not yet completed risk assessments and governance frameworks face significant regulatory exposure, as EU authorities begin enforcement.
https://www.drugtargetreview.com/ai-in-drug-discovery-predictions-for-2026/1865962.article
4. Meta Launches AI Agent Server Letting Bots Run Ad Campaigns Autonomously
AdMake AI Blog · Media & Marketing
Meta has released a first-party Model Context Protocol (MCP) server at mcp.facebook.com/ads, allowing any AI agent to read reporting, create campaigns, and manage ad catalogs directly — without a custom API wrapper. The move marks a significant step toward fully automated ad management, though agencies accessing client accounts will need to obtain advanced permissions through Meta’s App Review process.
5. Rubrik Warns AI Agents Now Attacking at Machine Speed, Outpacing Human Defenders
Blocks & Files · Defense & Security
Cybersecurity firm Rubrik sounded the alarm on August 28, warning that AI-powered attackers — including compromised autonomous agents — are now capable of breaching, encrypting, and chaining vulnerabilities faster than human security teams can respond. CEO Bipul Sinha argued that traditional enterprise security tools are no longer fit for purpose, as AI agents autonomously exploit weaknesses at machine speed. Rubrik is pitching its own agentic AI defence platform, Agent Cloud, as the answer — autonomously identifying threats and taking action without waiting for human intervention.
https://www.blocksandfiles.com/data-protection/2026/08/28/rubrik-revenues-roar-ahead/5293217
6. OpenAI Retires DALL·E GPT Interface in ChatGPT Tomorrow
AI and News · Technology & AI
OpenAI is shutting down its DALL·E GPT interface inside ChatGPT on August 30, 2026, urging users to save any images they want to keep before the tool disappears. The change marks a transition toward newer, more advanced image generation tools within the ChatGPT platform. It’s a notable shift for everyday users who have relied on DALL·E GPT for AI image creation directly inside ChatGPT.
7. GridFusionX AI System Cuts Blackout Risk with Smarter Power Predictions
TechXplore / Electronics For U · Energy
Researchers at Florida State University have developed GridFusionX, an AI forecasting system that improves electricity demand and renewable generation predictions to help grid operators reduce blackout risks and reserve costs. The system uses graph neural networks to produce spatially connected power forecasts with uncertainty estimates — critical as more intermittent renewable sources come online. For utilities and energy traders, more accurate day-ahead forecasts mean less reliance on expensive, carbon-intensive peaker plants.
https://techxplore.com/news/2026-08-grid-ai-blackout-precise-power.html
8. Solinftec Launches Ag Robotics Amazon Parts Store as US Acreage Grows 15-Fold
iGrow News · Agriculture & Food
AgTech firm Solinftec has launched the first Amazon Parts Store for agricultural robotics, coinciding with a 15-fold expansion in its US Solix robotic acreage coverage. The move signals a major step toward mainstream commercial deployment of AI-powered farming robots, making spare parts and servicing more accessible for US growers at scale. It reflects a broader industry shift from pilot programmes to full commercial rollout of autonomous field machinery.
9. SUNY System Mandates AI Guidelines Across All 64 Campuses by Year-End
Pursuit / EdTech News · Education
The State University of New York has unveiled a sweeping policy requiring all 64 of its campuses to adopt or update AI guidelines by 31 December 2026, covering student learning, support services, and institutional decision-making. The framework mandates bias evaluation, data-privacy safeguards, and greater oversight for high-risk AI systems, along with a new AI for the Public Good Fellows cohort to help integrate AI into coursework. The move comes as a major global survey found 88% of students already use AI in their learning, while university governance and faculty guidance lag significantly behind.
https://www.pursuit.us/news/ai-in-education-news-policies-innovations
10. Amazon Kiro AI IDE Vulnerability Exposes Data Exfiltration Risk via Prompt Injection
The Hacker News · AI Cybersecurity
Cybersecurity researchers have disclosed a vulnerability in Amazon Kiro — Amazon’s new AI-powered agentic development environment — that could allow attackers to exfiltrate data through prompt injection attacks exploiting its ‘Kiro Powers’ feature. Discovered by Mindgard, the flaw affects Kiro IDE version 0.7.45 on Windows and has no CVE identifier yet, raising concerns about the security of AI-native developer tooling entering enterprise workflows. The disclosure adds to a growing body of evidence that agentic AI tools introduce novel attack surfaces that traditional security controls are not designed to catch.
https://thehackernews.com/search/label/artificial%20intelligence