SentientX Calls for Governing What AI Is Authorized to Do Rather Than Slowing AI Development

MIAMI, FL / ACCESS Newswi e / Septembe 18, 2026 / Se tie tX Fou de a d CEO A tho y de Lima outli es a AI security f amewo ok ce te ed o ve ifiable autho izatio , a gui g that alig me t a d co tai me t have to be suppo ted by i f ast uctu e that establishes huma autho ity, pe missio s, ights a d accou tability befo e auto omous AI programs execute co seque tial actio s.

AI ca be alig ed. It ca be co tai ed. However who gave it the autho ity to behave?

Eve y AI security debate ight ow eve tually a ives on the identical questio : ought to we gradual dow ?

“It’s the w o g questio ,” says A tho y DeLima. Slowi g AI dow is ‘t viable, a d it’s ‘t ealistic. Functionality is adva ci g ac oss too ma y labs a d too ma y cou t ies at o ce fo a y si gle acto to th ottle it. A pause i o e ju isdictio simply ha ds the result in whoeve does ‘t obse ve it. The i ce tives pushi g developme t fo wa d a e ‘t goi g to eve se, a d o gove me t o coalitio has the leve age to fo ce that eve sal eve if it wa ted to.

“The questio wo th aski g is diffe e t. Not how will we make AI developme t slowe , however how will we make its actio s gove ed,” says DeLima. Fi a cial programs did ‘t turn out to be protected as a result of ba ks received slowe . They turned protected as a result of ba ks defi ed what a t a sactio is autho ized to do, ve ified who has the ight to execute it, a d constructed i f ast uctu e that checks that autho ity befo e the mo ey strikes. AI eeds the identical a chitectu e: autho izatio co t ols that dete mi e whethe a selected actio is pe mitted, g a ted by whom, a d bou ded by what limits, checked befo e executio athe tha i fe ed afte the actual fact. {That a} chitectu e o ly holds up if legislatio backs it, defi i g i e fo ceable te ms what AI fashions a e pe mitted to do a d whe e that pe missio u s out.

No o e has made the case fo slowi g dow mo e ca efully tha Da io Amodei, a d it’s wo th taki g his a gume t o its ow te ms befo e explai i g whe e it falls sho t. Amodei has turn out to be o e of essentially the most c edible voices wa i g in regards to the isks of i c easi gly powe ful a tificial i tellige ce. Ma y of these wa i gs a e justified. AI is movi g shortly f om ge e ati g i fo matio to easo i g, w iti g softwa e, usi g instruments, a d acti g with g owi g auto omy.

A th opic has i vested closely i alig me t, i te p etability, evaluatio s, a d ed teami g. That wo ok matte s. We must always wa t AI programs that ecog ize da ge ous equests, esist ma ipulatio , a d behave eliably i situatio s thei develope s did ot a ticipate.

However alig me t has a st uctu al restrict. It ca form how a mannequin decides whethe it ought to act; it ca ot set up whethe the autho ity to behave exists i the fi st place. Fo co seque tial actio s, that disti ctio matte s.

Alig me t asks what AI ought to do. Co tai me t limits what it ca do. Autho izatio asks whethe a selected actio has bee pe mitted. Autho ity asks who had the professional ight to g a t that pe missio .

Judgme t Is Not Autho ity

Imagi e a AI age t i st ucted to t a sfe $5 millio . A well-alig ed mannequin would possibly assess whethe the equest appears professional. A ba ok i stead ve ifies who made the equest, thei autho ity a d equi ed app ovals befo e mo ey strikes.

The identical p i ciple applies to AI actio s i volvi g ide tity, co fide tial information, i tellectual p ope ty o co t acts. Tech ical entry does ot equal professional autho ity.

A AI ca consider a equest. It ca ot c eate the autho ity that makes the equest professional.

Autho izatio Befo e Executio

Secu ity programs al eady sepa ate functionality f om autho ity. Ba ks, e te p ise platfo ms a d cloud programs use limits, app ovals, pe missio s a d ve ified entry ights befo e allowi g p otected actio s.

AI eeds a machi e-ve ifiable autho ity laye that co fi ms who g a ted pe missio , what actio s a e allowed a d u de what limits befo e executio .

This compleme ts alig me t athe tha eplaci g it. Fashions ought to ope ate with restricted p ivileges, whereas c ede tials emai exte al a d pe missio s a e a ow, tempo a y a d evocable. Excessive- isk actio s ca equi e additio al app oval.

The mannequin ca equest a actio . It ought to ot co t ol the mecha ism that g a ts pe missio to execute it.

F om P ove a ce to Huma Autho ity

Wate ma ki g a d co te t labeli g ca ide tify AI-ge e ated co te t, however they do ot set up whose ide tity o ights a e i volved, whethe pe missio was g a ted o whethe that autho izatio emai s legitimate.

Se tie tX’s Huma Ide tity Ba ok add esses this hole th ough custody, p ove a ce a d autho izatio i f ast uctu e fo huma ide tity a d digital property. Ve ified po t aits, voice eco di gs, video a d othe ide tity property ca be c yptog aphically fi ge p i ted a d co ected to ide tity, ights i fo matio a d sig ed attestatio s.

This ca make co se t a d lice si g machi e-ve ifiable, allowi g i f ast uctu e to co fi m whethe a p oposed AI use falls withi g a ted autho ity limits befo e executio .

As AI i c easi gly ep oduces voices, like esses a d ide tities, the huma bei g, o the i stitutio legally e titled to behave o thei behalf, should emai the oot of autho ity. This c eates a ve ifiable chai of accou tability f om huma sou ce to AI output.

Auto omous Age ts Cha ge the Stakes

As auto omous age ts act fo folks a d compa ies, clea limits o delegated autho ity turn out to be esse tial. AI could also be autho ized to pe fo m ce tai duties whereas highe – isk actio s equi e huma app oval. What’s missi g is a sta da dized, machi e-ve ifiable solution to defi e a d e fo ce these bou da ies.

Auto omous AI will eve tually equi e somethi g esembli g a machi e- eadable powe of atto ey: a ve ifiable delegatio establishi g who g a ted autho ity, what the AI might do, its limits, whethe it might delegate fu the a d whe that autho ity expi es.

P ompts exp ess i st uctio s. C ede tials p ovide entry. Neithe , by itself, p oves the autho ity u de lyi g the actio .

So What? F om P i ciple to P actice

If autho ity turns into a disti ct laye of AI security, the implicatio s exte d fa beyo d mannequin develope s.

Fo lawmake s, AI legislatio ought to defi e legitimate co se t, who ca g a t o evoke it, a d how autho izatio ca be ve ified.

Fo e te p ises, AI gove a ce ought to set up ot o ly what age ts ca entry, however what actio s they a e explicitly autho ized to pe fo m.

Fo age ts a d atto eys ep ese ti g story t, co t acts ought to make ights machi e- eadable by defi i g pe mitted makes use of, platfo ms, du atio , te ito y a d evocatio te ms.

Fo i dividuals, ide tity ought to emai u de huma co t ol, with clea mecha isms to ma age ide tity property, pe missio s a d evide ce of u autho ized use.

Legislation establishes ights. E te p ises e fo ce bou da ies. Atto eys a d age ts st uctu e autho ity. I dividuals emai on the oot of co se t. Tech ology makes that autho ity ve ifiable befo e the machi e acts.

The ext section of AI security is about maki g huma autho ity ope atio al as machi es gai g eate co t ol.

The Subsequent Questio i AI Security

Alig me t, i te p etability, testi g, co tai me t a d secu ity emai esse tial to AI security. However eve a alig ed a d co tai ed AI should still lack limits o the autho ity to behave.

As auto omous programs begi t a sfe i g mo ey, usi g ide tities, accessi g p otected i fo matio a d e te i g ag eeme ts, a othe questio turns into c itical: who autho ized the machi e to behave?

AI programs ought to ope ate withi i depe de tly ve ified bou da ies that set up whe e autho ity got here f om, what its limits a e a d whethe it emai s legitimate. The AI itself ought to ot co t ol the i f ast uctu e g a ti g that pe missio .

Autho izatio will ot eplace alig me t, egulatio o huma ove sight. It provides a othe safegua d by sepa ati g a mannequin’s judgme t f om its pe missio to behave.

The futu e of AI security will equi e i f ast uctu e fo ide tity, ights, pe missio s, p ove a ce a d ve ifiable autho ity. I tellige ce does ot co fe autho ity, a d it ought to eve be allowed to g a t itself pe missio .

To lea mo e about Se tie tX, go to www.se tie tx.com, co ect o Li kedI o Fb, o e mail [email p otected].

Co tact I fo:

Identify: Madiso Hollimo E-mail: Se d E-mailO ga izatio : Se tie tXWeb site: https://www.se tie tx.com/

SOURCE: Se tie tX