FTI Consulting News Bytes – 20 March 2026

FTI Consulting News Bytes

This week, Rachel Reeves unveiled plans for a £2.5bn UK government investment in quantum computing and AI, underscoring a drive to keep Britain competitive in the global technology race. We then turn to developments inside the workplace, where employees report rising cognitive strain linked to the expanding suite of AI tools. AI safety and governance also come into focus, as both Anthropic and OpenAI are looking to recruit experts to guard against the “catastrophic misuse” of their models. Legal scrutiny intensified, with Encyclopedia Britannica and Merriam-Webster suing OpenAI over alleged large-scale copyright infringement in a case that could help define how copyrighted material may be used to train AI systems. Finally, Instagram has confirmed that it will remove end-to-end encryption from direct messages in 2026, giving the company access to its users’ ‘private’ content.

This week’s news

Reeves invests in Britain’s tech future

Chancellor Rachel Reeves is placing advanced technologies at the centre of her growth strategy, with a particular focus on quantum computing as a key driver of future economic expansion. The BBC reports the government will invest £2.5bn in AI and quantum to accelerate adoption and prevent UK tech firms and talent from “drifting abroad.” According to the Financial Times this promise includes a £1bn commitment to purchase UK-built quantum computers in the 2030s, signalling confidence in the technology’s long-term potential. Together, the measures position quantum not just as an emerging field, but as a foundational technology for future growth, with public investment aimed at anchoring the industry in the UK and securing leadership in next-generation computing.

The hidden cost of workplace automation

The rise of workplace AI is contributing to a new form of stress dubbed “AI brain fry,” characterised by mental fatigue, reduced focus and a sense of diminished productivity. The Telegraph reports that while tools like ChatGPT promise efficiency gains, many employees are instead working longer hours and making more mistakes as they juggle multiple AI systems. Research cited from Boston Consulting Group found heavy AI users were less efficient and made significantly more errors, while ActivTrak data points to reduced time spent on focused work. Experts argue that rather than easing workloads, AI raises expectations and encourages constant multitasking, intensifying pressure. While some users benefit from automation, for many the need to oversee AI tools is adding to cognitive overload and workplace anxiety.

Anthropic seeks weapons expert to prevent AI misuse

Anthropic is hiring an external expert in chemical weapons and high yield explosives to help prevent “catastrophic misuse” of its AI systems, reflecting concerns that advanced models could be exploited to generate instructions for chemical, radiological, or explosive weapons. The BBC reports that the posting requires applicants to have at least five years of experience in chemical weapons or explosives defence, along with expertise in radiological dispersal devices. OpenAI is taking a similar approach, advertising a role in “biological and chemical risks.” However, Dr Stephanie Hare, a technology researcher and co-presenter of the BBC’s AI Decoded, has warned of the risks of AI systems handling sensitive, weapons-related information, noting that regulation in this area remains limited and underdeveloped.

The dictionary sues OpenAI 

Both Encyclopedia Britannica and Merriam‑Webster have sued OpenAI for copyright infringement, alleging the company scraped nearly 100,000 copyrighted Britannica articles to train its language models without permission. TechCrunch reveals the companies’ claims that ChatGPT sometimes produces “full or partial verbatim reproductions” of Britannica content and improperly uses the publisher’s articles in its retrieval augmented generation system. Britannica further accuses OpenAI of violating the Lanham Act, arguing that ChatGPT fabricates information and falsely attributes it to the publisher, jeopardising “the public’s continued access to high‑quality and trustworthy online information.” The complaint adds that ChatGPT directly competes with Britannica by generating answers that reduce the need to visit its sites. With no definitive legal precedent on whether training AI on copyrighted material counts as infringement, the outcome of this case may prove highly symbolic and could help define the boundaries of AI copyright law.

Instagram to end encrypted messaging

Instagram will remove end‑to‑end encryption of its direct messages in May 2026, according to The Guardian, allowing Meta to view all private message content. Meta says the change is due to very low uptake of the optional encryption feature, however, the decision follows years of pressure from law enforcement and child safety groups, who have argued that encryption makes it harder to detect exploitation and other serious harms. While regulators recognise the privacy benefits of encryption, they stress that platforms must still prevent harm. Digital Rights Watch’s Tom Sulston believes the move is less about law‑enforcement pressure and more about Meta shifting strategy, separating social media messaging from encrypted chat services like WhatsApp, which will remain fully encrypted. He also notes that removing encryption likely factored financial consideration as it could open the door for Meta to use message content for advertising or AI training, given the commercial pressure. Sulston argues that companies should be expanding, not reducing, end‑to‑end encryption.

Top Tweets of the Week

• Chris Stokel-Walker, tech journalist and lecturer, tweets: “The UK is reportedly championing government departments buy British tech to wean itself from US dependency. For @Independent I explain why I’m… sceptical of that.”
• Interesting Engineering, a page focussed on cutting-edge technologies, tweets: “Researchers have taught a humanoid robot to play tennis in only 5 hours. This represents a significant milestone for enabling robots to learn new skills quicker.”
• Tech Crunch tweets: “A robotic Olaf from Frozen brings #NVIDIAGTC’s keynote with CEO Jensen Huang to a close, showcasing a partnership between Nvidia, Disney, and DeepMind AI.”

Number of the week

$1tn Jensen Huang’s prediction for Nvidia’s AI chip revenue over the next two years  (Financial Times).