Ordering 212 takeaways in a single year would give most people pause for thought. But for one wealthy sports star, the calorific habit may prove to be particularly costly – at least if the tax authorities can prove their case.
The 212 Deliveroo orders in question are being cited by French tax authorities as part of their challenge as to the residence claims of Samir Nasri, a retired professional footballer living between Dubai and Paris.
Nasri – who represented Arsenal, Manchester City and the French national team as a player – claims to be resident in the UAE (and thus exempt from French taxes), but the authorities in France say the scores of takeaway receipts tell a different story.
Related
The Nasri case isn’t the only high-profile example of tax investigators digging deep into the weeds to contest residency claims. The Colombian pop star Shakira has been engaged in a similarly bitter dispute with Spanish tax authorities, which have used Instagram posts and credit card receipts to challenge Shakira’s contention that she was resident in the tax-friendly Bahamas in the early 2010s.
Traditionally, this kind of forensic examination of documents, receipts and other evidence has been relatively rare, and reserved for only the most high-profile cases. But the advent of AI raises the prospect that a similar level of scrutiny could be applied far more widely.
HMRC, the British tax administrator, has made a great fanfare of having appointed its first chief AI officer, as well as inking a partnership with Microsoft as part of a broader digital transformation. The Revenue says it will ‘close the tax gap’ by ensuring that individuals and businesses alike pay what is owed.
[See also: Is Britain’s wealth exodus about to accelerate?]
Content from our partners
The strategy may already be bearing fruit. HMRC’s investment in tax investigations is returning more money than ever, according to analysis by the law firm Pinsent Masons. Of every pound the tax man spends on investigations, the HMRC recovered £35 in unpaid tax last year – a rate of return that rose 13 per cent on the previous year. The return on investment ‘shows why the Chancellor has consistently awarded additional funding to HMRC’, notes the law firm. ‘The organisation continues to expand compliance activities, including investing in technology.’
While AI hasn’t become the all-seeing eye just yet, it is helping HMRC pursue several strategic goals, says Ian Robotham, tax partner with Pinsent Masons. ‘The first is helping HMRC join the dots between the different datasets that it has, including third-party data,’ he says. ‘The second is using AI to review that data to help to identify risk. And the third is aiding investigators to act on intelligence more efficiently in their active casework.’
The effect on individual cases will vary. But the overall effect, at least when applied across the organisation, is expected to be transformative. As more tasks are automated, time saved can be spent elsewhere. ‘You’ll likely see more investigations; more novel areas being challenged; and more compliance obligations,’ Robotham tells Spear’s.
[See also: How Budget 2026 could affect your wealth: what the UK’s top advisers are saying]
The already vast and quickly expanding volume of data available to tax authorities is a key part of the equation, as Nasri’s apparent Deliveroo habit attests. ‘These days, almost every transaction you make leaves a long-lasting digital footprint,’ says Matt Greene, a tax disputes partner at law firm Stewarts. With AI, that data can be processed in seconds.
But AI also makes mistakes. In another case relating to the French tax authorities, an AI agent tasked with scouring aerial photographs for evidence of swimming pools ended up falsely labelling a large blue gazebo as an undeclared pool.
The technology can also jump to premature conclusions when working with limited datasets. ‘We were acting in respect of a residency enquiry and HMRC was clearly using information obtained from border control records as to when the individual had entered the UK,’ says Jeff Millington, a former HMRC investigator and now tax adviser with the law firm Withers. ‘What they had not taken into account, though, was the times our client left the UK on a private plane.’
[See also: Dan Neidle: The rich can still silence their critics with lies]
For those familiar with HNW tax cases, the tightening of the net is nothing new.
‘I remember having an investigator years ago who was looking at a client’s flight times and meeting locations in an attempt to ascertain whether they had really made those trips,’ says Greene. Back then, such an investigation may have taken a determined desk officer weeks to complete. These days, with the right data, it could be done in minutes.
For now, HMRC insists it will not be using AI to make final decisions. It has also said that some of the more invasive tactics, such as using AI to interrogate a person’s social media or online mentions, will be reserved for criminal investigations rather than routine inquiries. But the overall picture seems clear: more sophisticated use of data, increased investment in investigations – and more scope for complex disputes, with or without reams of Deliveroo receipts.
