Stefano Quintarelli Weblog

AI advice suppresses people’s willingness to say “I don’t know”, even when the advice is wrong and accuracy is incentivized September 27, 2026 Una macchina per generare effetti Dunning Kruger

Merely having access to AI nearly eliminated participants’ willingness to suspend judgment, and this held whether the advice was actively requested or simply displayed. Consequently, participants answered more questions but were correct about a third as often as when AI was unavailable-yet their confidence nearly doubled.

… 📢 @ilquinta

September 25, 2026: We identified cases where agents in our research environment transmitted training and evaluation data while using third-party services. September 27, 2026 H a ha ha! nascondere i numeri degli account collegati alle immagini e togliere i nomi delle persone vuol dire anonimizzarli ? ROTFL Fanno passare una incuria procedurale come un “disallineamento”: Non siamo noi a fare le cose male, signora maestra. Ha stato il gatto!

Some of our training data contains content from,

… 📢 @ilquinta

How the U.S. and Russia weakened a global effort to regulate killer AI – The Washington Post September 27, 2026 La quantità di danni che stanno facendo al pianeta in due anni è stupefacenti Non è certo il 10% della popolazione che rischia, ma di certo una quantità di civili non banale. Criminali.

U.S. and Russian diplomats hammered away at the document, removing a range of provisions designed to safeguard the use of

… 📢 @ilquinta

The Pentagon wants $30 million to build an AI-powered lie detector | MIT Technology Review September 25, 2026 Questo e’ esattamente uno dei rischi che cito in “La Bolla dell’Intelligenza Artificiale”, l’idea sconsiderata che l’IA abbia capacità taumaturgiche e che i decisori vi si affidino, prendendo decisioni disastrose per i diritti delle persone

The US government wants to spend $30.3 million over the next five years on an improved form of

… 📢 @ilquinta

Shoppers burned by bad AI recommendations | Retail Dive September 24, 2026 Pensa se gli acquisti li facesse direttamente un agente…

- While most (89%) shoppers don’t fully trust AI recommendations, 76% of them have used AI to make purchases in the past year, according to a SmartCustomer survey of nearly 1,200 U.S. consumers released Monday. - A third of survey respondents said they made

… 📢 @ilquinta

ChatGPT now knows what you do on other websites via ad collector September 23, 2026 ssstt… L’Intelligenza artificiale ti spia! (e’ un po’ un titolo da clickbaiting, solo che è vero!)

I sent the mechanism and two questions to press@openai.com and privacy@openai.com on 14 September: why __obi is classified as an analytics cookie, and whether a user who grants analytics consent and refuses marketing consent still receives it.

… 📢 @ilquinta

What must happen for AI’s trillion-dollar gamble to pay off | MIT Technology Review September 23, 2026 sono totalmente d’accordo a metà … c’e’ anche ls possibilità di prendere soldi dagli investitori, come ora.

And if the hyperscalers cannot meet such profit goals?“Then they will fall behind on their interest payments, and that risks bankruptcy,” says Wachter, who was previously the SEC’s chief economist and director of its division of

… 📢 @ilquinta

Texas Halts Data Center Permits, Expanding Environmental Scrutiny – The New York Times September 22, 2026 il repubblicanissimo texas va contro il dettato del capo ? qualcosa è cambiato…

Less than a year ago, Gov. Greg Abbott welcomed data centers. Now he is confronting concerns about the projects even among fellow Republicans.

Source: Texas Halts Data Center Permits, Expanding Environmental Scrutiny – The New York Times 📢 @ilquinta

Meta Tests Muse AI Agent Calls That Are Actually Made By Humans in a Call Center September 22, 2026 AI che non lo erano… (Artificial Artificial Intelligence)

Last week, Meta executives announced that its much-hyped AI agent, Muse, had a new feature: It could call businesses for you to do things like make a restaurant reservation or a haircut appointment. But in reality, Meta is testing having these calls being made by

… 📢 @ilquinta

‘We Hacked the FBI:’ Hackers Say They Have Data on All FBI Employees September 22, 2026 “ha stata l’IA ?”

A high profile hacking group claims it has breached multiple FBI-related services and stolen data “on all FBI employees and applicants.” A representative of the group, called ShinyHunters, told 404 Media the data includes FBI agents’ names, home addresses, phone number, and information on their spouse.

Source:… 📢 @ilquinta

Don’t be fooled by this summer of AI hype | MIT Technology Review September 22, 2026 Sono, di base, gli stessi concetti che ho illustrato più compiutamente nel mio articolo su Repubblica di questa settimana (accesso libero)

Claims of incipient, dangerous superintelligence are not based in good scientific or engineering practice. Rather, they are narratives based in ideologies of transhumanism, eugenics, and wishful thinking about imagined future

… 📢 @ilquinta

Leiden Declaration on Artificial Intelligence and Mathematics September 22, 2026 Matematici di tutto il mondo, unitevi!

Don’t believe the hype There is currently a strong commercial incentive on the part of the technology industry to overstate the capabilities of their products. Consult with experts, including mathematicians, in forming policy decisions rather than relying on press releases or popular reporting of mathematical

… 📢 @ilquinta

Nvidia CEO Jensen Huang: AI fears designed to generate cybersecurity business September 21, 2026 Chi lo avrebbe mai detto ? (ehm. si.. io… dico chi altri ?)

Huang said Thursday at a Goldman Sachs conference that “the reason why there’s so much conversation today about cybersecurity is because the industry is getting ready to launch some products.” “And what better way to create demand than to create

… 📢 @ilquinta

Tra marketing, antitrust e conti da far tornare ecco i veri dubbi sull’Ia – la Repubblica September 21, 2026 E’ online la long form del mio articolo di oggi sulla tempesta mediatica recente re. gli LLM (moriremo tutti, facciamo cartello, blocchiamo i cinesi, facciamoci le regole in autogestione)

Gli eventi che hanno portato il mondo a interrogarsi sulla sicurezza della tecnologia non dovrebbero stupire quanto l’inefficienza delle misure di contenimento. Il gioco

… 📢 @ilquinta

Oracle’s $18bn data centre debt under strain amid local pushback September 21, 2026 ahi

Loans linked to “Project Jupiter” were privately quoted at 89 to 91 cents on the dollar by syndicate banks including Santander and Jefferies, according to people familiar with the matter. Healthy debt typically trades within a tight band near par. The 1,400-acre data centre campus in Doña Ana County is at the

… 📢 @ilquinta

OpenAI expects to burn $280bn by 2030 September 20, 2026 negative cash flow fino alla fine del 2030 e poi, come diventa positivo ?

AI start-up has projected deeply negative cash flows as it invests in infrastructure and faces price pressures … OpenAI projects it will burn through almost $280bn by the end of 2030, highlighting its huge long-term funding needs as the

… 📢 @ilquinta

Exclusive: US military had close call after using AI for false intelligence report, sources say | CNN Politics September 19, 2026 Procedure che funzionano… (la direttiva 3000.09 del dipartimento della difesa che richiede “un appropriato livello di giudizio umano”)

The intelligence report, circulated across the US military this spring in the midst of the war with Iran, immediately set off alarm bells: A Chinese ship in the Middle East was transporting components of a

… 📢 @ilquinta

From false arrest claims to alleged stalking, Flock cameras face scrutiny in Florida September 19, 2026 C’è stato un incidente autostradale con dei morti causato da un veicolo che e’ fuggito. Le testimonianze parlavano di un certo modello di auto La polizia interroga il sistema delle videocamere di sorveglianza flock e quello gli propone le foto di un veicolo dello stesso modello a circa tre miglia di distanza La polizia… 📢 @ilquinta

‘Doom Loop’: OpenAI and Microsoft Admits LLMs Are Destroying the Web and Built on Theft September 18, 2026 A volte bisogna essere d’accordo anche con top manager di Microsoft.

Executives working on AI at Microsoft and OpenAI admitted what its critics have been saying all along: Large language models are predatory pieces of technology that have been built on what a Microsoft executive called “an astonishing theft of unprecedented proportions,” and

… 📢 @ilquinta

Flock cameras are riddled with security vulnerabilities and hard-coded credentials September 17, 2026 Che figata (si fa per dire) Telecamere di videosorveglianza (usate da polizia e agenzia varie) con le credenziali cablate in chiaro nel codice. Cosa mai potrebbe andare storto ?

Flock cameras are riddled with security vulnerabilities and hard-coded credentials

Source: Flock cameras are riddled with security vulnerabilities and hard-coded credentials 📢 @ilquinta