AI News Digest: 24 March 2026

24 March 2026

Quick Read: Today's highlights cover the key AI developments from 24 March 2026, including the most important stories for UK businesses and decision-makers.

A day of bold claims and quiet power grabs. Nvidia's Jensen Huang declares AGI has arrived, MPs scramble to halt Palantir's access to Britain's financial watchdog, and new research shows UK businesses are racing ahead with AI adoption while leaving governance in the dust. Meanwhile, OpenAI and Anthropic wage an unprecedented bidding war for private equity backing.

MPs Demand Halt to Palantir's Access to FCA Regulatory Data

Cross-party MPs have written to the government urging it to halt a contract giving US data analytics firm Palantir access to the Financial Conduct Authority's internal "data lake". The three-month trial, worth more than £30,000 per week, grants Palantir access to a sprawling repository of regulatory intelligence covering fraud, money laundering, insider trading, and consumer complaints.

The FCA defended the deal before lawmakers today, insisting Palantir will act strictly as a "data processor" with no access to regulatory intelligence, and must destroy all data on contract completion. But critics see a familiar "land and expand" pattern: Palantir has already racked up over £500 million in UK public sector contracts across the NHS, policing, and defence.

Our take: This matters for every UK business handling sensitive data. The tension between sovereign capability and dependency on a handful of US tech vendors is becoming a defining policy question. If you are outsourcing data processing to third parties, watch how this plays out. The governance standards being demanded of Palantir today will likely shape expectations for all AI data processors tomorrow.

Half of UK Businesses Now Use AI, But Governance Lags Dangerously Behind

Gallagher's third annual AI adoption survey paints a striking picture: 50% of UK businesses have now implemented AI in some form, up from 20% in 2023 and 39% in 2024. An overwhelming 93% of UK leaders see AI as a positive opportunity, and 86% report a positive revenue impact.

The standout finding, however, is the governance gap. Adoption is surging through chatbots (52%), research and analytics (49%), Customer service (48%), and fraud detection (44%), but formal AI governance frameworks remain thin on the ground. UK businesses expect it will take around 27 months for AI investments to deliver clearly measurable financial returns.

Our take: These numbers confirm what we see on the ground at Precise Impact. UK businesses are not waiting for permission to adopt AI. But the gap between adoption and governance is a ticking clock. If your organisation is in that 50% using AI without a formal framework, the risk is not hypothetical. Get your AI governance strategy sorted before a regulator or a data breach forces the issue.

Jensen Huang Declares "We've Achieved AGI"

Nvidia CEO Jensen Huang made headlines on the Lex Fridman podcast by stating, "I think we've achieved AGI." Fridman's definition of AGI centres on an AI system capable of starting, growing, and running a billion-dollar company. Huang agreed that threshold has been met, pointing to platforms like OpenClaw and their growing real-world usage as evidence.

He then partly walked the statement back, acknowledging that while individual AI agents can accomplish remarkable tasks, the probability of them consistently running complex organisations remains low. The comment has sparked intense debate across the industry.

Our take: Strip away the hype and there is a genuine signal here. AI can now handle tasks that would have required entire teams two years ago. But "AGI" remains a marketing term as much as a technical one, and Huang's own caveats reveal that. For UK businesses, the practical question is not whether AGI exists. It is whether you are using the AI capabilities that already do exist to their fullest potential. Most organisations are not even close.

OpenAI Offers PE Firms 17.5% Guaranteed Returns in Enterprise Turf War with Anthropic

OpenAI is sweetening its pitch to private equity firms with a guaranteed minimum return of 17.5% to join a joint venture targeting enterprise AI distribution. The deal, valued at approximately $10 billion with $4 billion in PE equity, puts OpenAI in direct competition with Anthropic, which is pursuing a smaller $1 billion venture without guaranteed returns.

TPG is expected to anchor the deal, with Advent International, Bain Capital, and Brookfield as co-founders. Participating firms also get early access to OpenAI's newest models. At least two PE firms have already walked away from both deals, citing concerns about economics and flexibility.

Our take: This is the enterprise AI land grab in full swing. OpenAI and Anthropic are not just competing for users. They are competing for distribution channels through PE portfolio companies. For UK businesses backed by private equity, expect your investors to start pushing specific AI vendor partnerships. That is not necessarily a bad thing, but make sure the choice serves your business strategy, not just your investor's deal terms.

UK Government Steps Back from AI Copyright Reform

The UK government has confirmed it will not extend the text and data mining (TDM) exception in copyright law to facilitate AI training, a move that diverges sharply from the EU's approach. The government previously expressed a preference for a broad copyright exception with an opt-out mechanism but has now reversed course, choosing to maintain a monitoring brief and gather more evidence.

This leaves the UK in a curious position: no new TDM exception for AI developers, but no strengthened protections for creators either. The EU, by contrast, already has broader TDM exceptions in place under Directive 2019/790.

Our take: For UK AI companies, this is a "wait and see" outcome that pleases nobody. Developers wanted clarity on training data rights. Creators wanted stronger protections. Instead, both get uncertainty. If your business relies on training models with third-party data, you still need robust licensing agreements. The legal risk has not gone away; it has simply gone unresolved.

Quick Hits

Frequently Asked Questions

Has AGI actually been achieved as Jensen Huang claims?

It depends entirely on how you define AGI. Huang agreed with Lex Fridman's definition of an AI that could run a billion-dollar company, then immediately added caveats about consistency and reliability. Most AI researchers would argue we are still some distance from artificial general intelligence in the traditional sense. What is true is that current AI systems can handle an impressive range of complex tasks. For businesses, the practical capabilities matter far more than the label.

Why are MPs concerned about Palantir's FCA contract?

The concern centres on a US company gaining access to highly sensitive UK financial regulatory data, including fraud cases, money laundering investigations, and consumer complaints. Critics see it as part of Palantir's pattern of embedding itself across UK public services, making it increasingly difficult to remove. The broader worry is about digital sovereignty and the UK's dependency on a small number of large American tech vendors for critical national infrastructure.

What does the UK AI copyright decision mean for businesses training AI models?

The UK government has decided not to extend text and data mining exceptions for AI training, but has not introduced new restrictions either. This means the legal position remains largely unchanged and uncertain. Businesses training AI models on third-party content should continue to secure proper licensing agreements and maintain detailed records of training data provenance. The EU has taken a different path with broader TDM exceptions, which may create competitive implications for UK-based AI developers.