AI product management · Industry brief

The EU AI Act enforcement timeline follows a four-phase rollout, with prohibitions and AI-literacy requirements already in effect as of February 2025, general-purpose AI obligations beginning August 2025, and core high-risk system obligations arriving August 2026. For deployers of high-risk systems—organisations using AI in regulated contexts—the 2026-2027 window introduces substantial operational requirements including human oversight, automatic logging, incident monitoring, and decision traceability. A proposed Commission amendment in late 2025 may defer some high-risk obligations beyond the statutory dates by linking them to the availability of harmonised standards, though this change takes legal effect only upon formal adoption and publication in the Official Journal, leaving the August 2026 and 2027 dates as the current operative planning assumptions.

North American venture investment reached record levels in the first half of 2026, totaling $392 billion, with Q2 alone accounting for $137.2 billion—the second-highest quarterly total on record. AI startups dominated funding, capturing approximately 80% of investment across all stages in Q2, with Anthropic securing the quarter's largest round at $65 billion (led by Altimeter Capital, Dragoneer, Greenoaks, and Sequoia Capital, plus corporate investments from Amazon and Google), followed by Prometheus at $12 billion. Anthropic subsequently filed confidentially for an IPO, while the quarter also saw SpaceX achieve the largest IPO of all time at $75 billion in June, and SpaceX completed the largest startup acquisition ever—a $60 billion purchase of AI coding tool Cursor from parent company Anysphere—after announcing the acquisition option in April.

The EU AI Act (Regulation 2024/1689) enters staged implementation with key compliance milestones: prohibited practices and governance obligations live as of August 2024; transparency rules and GPAI obligations activate August 2025; majority enforcement begins August 2026 with Article 50 transparency requirements; high-risk AI systems apply December 2027 for standalone systems and August 2028 for product-embedded systems, with public-authority legacy systems under transitional rules through August 2030. Boards must convert regulatory dates into AI inventory, disclosure architecture, vendor management, documentation, and evidence automation, with executive ownership assigned for compliance readiness across the 2026–2030 runway. By August 2026, enforcement readiness becomes the enforcement pivot point, requiring organizations to evidence defensible control systems for AI classification, high-risk obligations, and post-market monitoring.

OneTrust published guidance on October 20, 2025, establishing a formal AI Governance Committee framework with a structured buy-versus-build evaluation process for AI tools and three mandatory controls for agentic AI systems: restrictions on decision-making authority, full action traceability, and least-privilege data access. The framework applies globally and is positioning itself as a de facto industry benchmark that regulators and auditors are increasingly using to assess organizational maturity, creating potential audit exposure for enterprises without documented AI governance structures. Agentic AI controls are becoming a regulatory expectation across multiple jurisdictions, with frameworks from Singapore's IMDA, the EU AI Act, and emerging U.S. state laws all mandating traceability and autonomy restrictions. Regulators and standards bodies including the EU AI Office, Singapore's IMDA, and NIST are expected to publish more prescriptive guidance on autonomous system controls in 2026, while enforcement actions may begin citing inadequate oversight of autonomous AI as an aggravating factor and procurement-focused regulators in financial services and healthcare are tightening third-party AI tool intake requirements (Source: OneTrust governance framework publication, October 20, 2025).

Global venture funding reached record levels in early 2026, with Q1 attracting $285.5 billion—though OpenAI's $122 billion raise accounted for 43 percent of that total, according to CB Insights' "State of Venture Q1'26" report. The top five deals (OpenAI, Anthropic, xAI, Waymo, Databricks) represented approximately three-quarters of total venture investment, reflecting unprecedented concentration in AI-related companies. This bifurcated landscape sees mega-rounds in AI, semiconductors, and cloud infrastructure surging 77 percent to 738 deals capturing 65 percent of funding, while early-stage activity remains subdued and deal count fell 17 percent to 29,501 in 2025, per CB Insights' "State of Venture 2025" report. US-based companies dominated, raising $328 billion (70 percent of global funding) in 2025, while Asia grew 7 percent to $53 billion and Europe rose 18 percent to $68 billion. KPMG's "Q1'26 Venture Pulse Report" noted that defense tech and spacetech gained traction supported by government engagement, while M&A is expected to play an increasingly important role as an exit path amid ongoing market uncertainty.

Radware announced enhancements to its Agentic AI Protection solution, adding compliance reporting to support alignment with ISO 42001, the European Union AI Act, and the NIST AI Risk Management Framework. The release expands visibility into AI agent ecosystems through enhanced monitoring capabilities and introduces audit-ready reporting designed to help organizations demonstrate accountability and traceability across AI-driven workflows. Protection has also been extended to developer-hosted AI agents, including Anthropic Claude Code, as enterprises face growing regulatory pressure to demonstrate control and oversight of autonomous systems. Source: HelpNetSecurity (helpnetsecurity.com, published 2026-07-07)