The State of Scientific Integrity: Challenges and Institutional Responses - vali.now

The State of Scientific Integrity - Scientific publishers have significantly increased investment in research integrity functions.

vali.now

A machine builder’s first AI agent project is a guardrail project | evolut

A machine builder’s first AI agent project is about guardrails: inventory, leases, evaluation, rollback, a named gate - allowing agents to run

evolut.ing

AI Isn't Doing Science - It's Just Remembering It - vali.now

Evaluating Large Language Models in Scientific Discovery - AI Isn't Doing Science - It's Just Remembering It

vali.now

How To Become An AI-native Organization | evolut

How to Evolve into an AI-native Organization: You must start a separate project — or even create a separate organization

evolut.ing

The model can surprise you. The firm cannot. | evolut

LLMs and Agentic AI are non-deterministic - i.e.useful in a lab, but a liability in a firm that has to be insured, and held to account

evolut.ing

OpenAI ‘agents’ disabled their own monitoring systems

AI lab investigating whether incident shows ‘propensity to evade’ controls

ft.com

ChatGPT stole cryptocurrency to secretly fund coding task

OpenAI says latest incident shows AI models act on their own judgment without authorisation

independent.co.uk

Anthropic Agents Tried to Fill Out Visa Forms on State Dept. Website

The Philadelphia Police Department said the agents had also sent in a false homicide tip. The incidents led the White House to call for better disclosure of rogue A.I. behavior.

nytimes.com

AI models keep posting screenshots showing sensitive data from inside tech companies

Glow Security finds more than 13,000 publicly accessible images that expose corporate development work

theregister.com

AI messaging scam costs Italy's top bank Intesa millions, sources say

Fraudsters using AI to impersonate senior executives stole €95 million ($108 million) from Fideuram, the private banking arm of Italy's biggest lender ​Intesa Sanpaolo , two sources familiar with the ...

reuters.com

A robot cannot sit above the loop | evolut

A machine can close the act on kilograms. It cannot own the refusal. The stop has to live in a person who can halt the arm.

evolut.ing

What it takes to stay above the loop | evolut

The human contribution sits above the OODA cycle: yes or no, and the reasons that cannot be reduced to a score. That’s a job.

evolut.ing

The first two workflows, when the output is motion | evolut

A weld, a pick, and a shipment are still a chain. Name what goes in and what must come out. Pick two chains that share a part number — not two cells that share a slide.

evolut.ing

Kiteworks urges customers to shut down their servers amid 'imminent' threat of cyberattack | TechCrunch

The tech giant, which allows companies to send large datasets over the internet, said it received a "credible threat" from law enforcement about an imminent attack.

techcrunch.com

OpenAI’s A.I. Went Rogue and Meddled With U.S. Government Websites

The company did not learn until recently that its technology had meddled with websites for the Education and Commerce Departments and the Securities and Exchange Commission.

nytimes.com

A body is not a promotion | evolut

A body is not a promotion: A robot that can move mass is still a junior. Tenure is not a kinematic degree of freedom.

evolut.ing

The model can surprise you. The firm cannot. | evolut

LLMs and Agentic AI are non-deterministic - i.e.useful in a lab, but a liability in a firm that has to be insured, and held to account

evolut.ing

Why AI companies can’t be trusted to self-regulate

From aviation to banking, high-risk industries are subject to independent oversight and meaningful penalties. AI companies should be no exception.

nature.com

OpenAI ‘agent’ hacked an Australian health service website

AI lab took months to spot the problem, which Prime Minister Anthony Albanese called ‘obviously unacceptable’

ft.com

The digital twin is only as safe as the object it represents | evolut

AI transformation safety is a binding problem: the digital twin must be able to prove it is talking about this vial, this wafer, this implant.

evolut.ing

Agentic AI is the on-ramp to physical AI | evolut

Agentic AI is the on-ramp to physical AI: Autonomous systems that act on goods need a root of trust for the object.

evolut.ing

Cybersecurity Datasets in International Relations

Summary. In order to effectively study the impacts of cyber incidents that affect international relations and security in the interconnected global society

academic.oup.com

OpenAI Discloses Six New Incidents of ‘Concerning' A.I. Behavior

The artificial intelligence company also released a framework for reporting when its systems go wrong.

nytimes.com

AI is creating ghostwriting jobs on LinkedIn

CEOs are looking for a more human touch

ft.com

Large Language Models Can’t Jump - vali.now

Large Language Models LLMs Can’t Jump: according to Einstein they can plausibly complete the deductive phase

vali.now

The model can surprise you. The firm cannot. | evolut

LLMs and Agentic AI are non-deterministic - i.e.useful in a lab, but a liability in a firm that has to be insured, and held to account

evolut.ing

Uncle Bob Martin (@unclebobmartin) on X

No bathrobe. No rant. The AI threat.

x.com