Why AI could be a financial ‘sludge’ buster

Financial Times Europe11 Jul 2026Gil lian Tett gil lian.tett@ft.com

Why are so many people in Amer ica (and else where) so neg at ive on AI? There aremany reas ons: hor ror at tech com pan ies’ excess power; fears about impend ing joblosses; alarm about AI’s impact on men tal health, energy and the envir on ment.

Then there is a prac tical issue. While AI can the or et ic ally deliver mir acles — in lifesci ences, say — few people have yet seen much tan gible improve ment in their ownlives. Win-wins are prov ing elu sive.

Tech ies and gov ern ments are now try ing to address this pub lic rela tions prob lem.In the UK, for instance, Doug Gurr, former chair of the Alan Tur ing Insti tute, tellsme that the Met Office is revolu tion isingflood fore casts with AI, and that the Nat -ural His tory Museum can now identify dino saur bones 30,000 times faster. Jur as sicPark is optim ised.

But fin ance might soon pro duce another win-win too. As Michael Hsu, formerly oneof Amer ica’s top bank reg u lat ors, told the Cam bridge Digital Innov a tion and Reg u la -tion Ini ti at ive (which I co-con vene), AI could also slash reg u lat ory red tape. If so,that would not just help har assed reg u lat ors and bankers but con sumers of fin an -cial ser vices as well.

“AI is a force mul ti plier,” says Witit Syn satayakul of the Asian Devel op ment Bank,which has pub lished a potent recent paper on the topic.

The key issue is what Cass Sun stein, the beha vi oural eco nom ist, calls “sludge” — thefact that 21st-cen tury life is beset with red tape, gen er at ing “excess ive or unjus ti fiedfric tions, such as paper work bur dens, that cost time or money”. In 2015 alone,Amer ic ans spent 9.78bn hours on fed eral paper work.

That makes “life diffi  cult to nav ig ate” for cit izens because sludge is “frus trat ing,stig mat ising, or humi li at ing”, he notes, call ing for “firms, uni versit ies, and gov ern -ment agen cies [to] reg u larly con duct Sludge Audits” — and cut it.

That sounds sens ible. And groups like the Fin an cial Con duct Author ity are try ing:the FCA’s Helen Pack ard said on Thursday that the reg u lator will save £100mn forfirms by stream lin ing rules — even without AI. But this “sludge-bust ing” can be anight mare too. Hsu says that when the Reserve Bank of India tried to turn 9,445 cir -cu lars into a more rational 244 “mas ter dir ec tions”, it took almost 40 staff an entireyear. And when Con gress intro duced a law in 1996 requir ing US reg u lat ors to audit— and cut — rules every 10 years, it proved so “extraordin ar ily labour intens ive”that “the mere plan ning for the [review] pro cess” took over a year.The prob lem is multi-faceted. There is “ver tical sludge”, where top-down lawsbecome increas ingly com plex when turned into prac tical rules. And there is also“hori zontal slush”, where adja cent reg u lat ors cre ate over lap ping rules, and “timeslush”, when new rules are piled on top of old ones.

And now we have “geo pol it ical slush”: con flict and rising pro tec tion ism mean thatnational reg u lat ors are less eager to co-ordin ate laws. Just look at what Brexit hasdone or how US reg u lat ors reject new Basel bank ing rules.

But this is pre cisely why AI could help, given its abil ity to scan swell ing moun tainsof text, then syn thes ise and check it for incon sist en cies, gaps and over laps.

Stan ford Uni versity has already partnered with San Fran cisco city lead ers to pilotthis idea: a team recently stud ied muni cipal rules (totalling 16mn words), thenfocused on the legis lat ively man dated reports to find duplic a tions. The legis laturethen cut and con sol id ated 36 per cent of them. Yes, really. “Because of the length ofour code . . . it’s likely a project we would never have under taken [without AI],”notes David Chiu, city attor ney.

Now reg u lat ors want to rep lic ate that in fin ance. And it seems a good moment,given that insti tu tions like the European Cent ral Bank and the UK Treas ury facemount ing polit ical pres sure to slash red tape. Moreover, poli cy makers face an“arms race” in deal ing with bankers’ own AI innov a tions, a top UK reg u lator says.

So can this sludge-bust ing work? It won’t be easy, alas. One prob lem is that largelan guage mod els are bet ter at tack ling con ver sa tional or lit er ary texts than reg u lat -ors’ ghastly legalese. Another is the lim ited AI skills of reg u lat ors, while a third isthe danger the machines might go mad without a human in the loop, or imposeexcess ive stand ard isa tion.

But the tech nical prob lems are now being solved by new innov a tions. And Hsu saysnew legal tools such as Har vey, Irys, CoCoun sel, Norm AI and Kimi are out per form -ing main stream plat forms like Claude and Chat GPT for reg u lat ory tasks. Reg u lat orsare hir ing staff with AI skills and pledging to main tain allimport ant human over -sight.

Will that cut fin an cial rules by a third, as happened in San Fran cisco? It seems anambi tious dream. But any reduc tion would help. So three cheers for red-tape-redu -cing robots — they might yet make mar kets more effi cient and per haps encouraging.