Former Anthropic researcher Jason Coxon discusses his resignation from the company and concerns over artificial intelligence safety on ‘Special Report.’ #fox #media #breakingnews #us #usa #new #news #breaking #foxnews #specialreport #ai #artificialintelligence #anthropic #jasoncoxon #technology #aisafety #superintelligence #aitakeover #superweapon #futureofai #technews #innovation #security
Don’t just watch Fox News—be part of it. Become a Fox News Patriot today.
Subscribe to Fox News:
Watch more Fox News Video:
Watch Fox News Channel Live:
Download the Fox News app:
FOX News Channel (FNC) is a 24-hour all-encompassing news service delivering breaking news as well as political and business news. The number one network in cable, FNC has been the most-watched television news channel for 18 consecutive years. According to a 2020 Brand Keys Consumer Loyalty Engagement Index report, FOX News is the top brand in the country for morning and evening news coverage. A 2019 Suffolk University poll named FOX News as the most trusted source for television news or commentary, while a 2019 Brand Keys Emotion Engagement Analysis survey found that FOX News was the most trusted cable news brand. A 2017 Gallup/Knight Foundation survey also found that among Americans who could name an objective news source, FOX News was the top-cited outlet. Owned by FOX Corporation, FNC is available in nearly 90 million homes and dominates the cable news landscape, routinely notching the top ten programs in the genre.
Watch full episodes of your favorite shows
The Five:
Special Report with Bret Baier:
Jesse Watters Primetime:
Hannity:
The Ingraham Angle:
Gutfeld!:
Fox News @ Night:
Follow Fox News on Facebook:
Follow Fox News on X:
Follow Fox News on Instagram:
source
Tell me again how blk ppl are destroying the world?
Reason Extraterrestrials like the Pleiadians and Arcturians said Humans are heading to a crisis with our Civilization towards Destroying our selves ,Human Race has a severe imbalance too much technology and not enough sprituality. Extraterrestrials nolonger rely on artificial intelligence because of whats happening with Humans and they have developed intelligence greater than AI,computers and AI have been rendered obsolete by them for many thousands of years.Human Race is in danger of losing control of what they have unleashed, thats due to our primitive negative Neanderthal behavior, """MONEY,GREED,POWER,LYING,CONQUEST,OWNERSHIP etc etc and many many other negative behaviors.
😮
this guy is so funny. his opinion sucks. you cannot get everyone in this world to agree with you. what about Iran? Iran may be doing this and they're not going to give it up no matter what you try and put as a rule to running it
And yet they keep building them… everyone is so eager to embrace their own demise…IS THINKING IM THE LAST SANE PERSON ALIVE MAKE ME CRAZY ?!?!?!?!
We believe you HARRY.⚡✨, ok we will slow down on A.I.👍
“There is no fate but what we make for ourselves”
When people who actually helped build these systems start using words like “superweapon,” I listen. Elias Velin’s book hits very differently after hearing this.
😮
😮
1. HIGH GOVERNMENT DEBT POLITICAL POLARIZATION, GEOPOLITICAL, COMPETITION, & ECONOMIC PRESSURE
2. A WEAKER DOLLAR,BANK FAILURE,OR GOVERNMENT SHUTDOWN BY ITSELF
3. BANK: THERE HAVE BEEN FIVE U.S. BANK FAILURE IN 2026, ACCORDING TO RECENT REPORTING , BUT THAT IS NO WERE NR EVIDENCE OF A 2008 STYLE SYSTEMIC S BANKING COLLAPSE. FDIC CONTINOUSE TO MONITOR FAILURE
4. DEBT : THIS IS ONE OF THE BIGGEST STRUCTURAL PROBLEM US DEBT HAS CROSSED $ 40 MILLION HIGHER TREASURY YIELD, MEANS THE GOVERNMENT MUST PAY MORE TO FINANCE OF DEBIT
5. BOND MARKET ; THIS A PROBABLY THE AREA I'D WATCH' MOST CLISELY. THE 10 YEARS TREASURE DECENTLY APPROACH 5% WHILE THE 30 YEAR YEILD REACHED AROUND 5.3% SUSTAIN YEILDS ABOVE 5% COULD PUT SIGNIFICANT PRESSURE ON STOCK, HOUSINGS, CORPORATION & GOVERNMENT FINANCE
6. A MAJOR FALL IN THE DOLLARS COMBINED WITH FOREIGN INVESTOR REDUCING TREASURY PURCHASE
7. THE COMBINATION OF HIGH DEBT – HIGH INTEREST RATES + INFLATION + WEAKER TREASURY DEMAND IS THE REAL STRUCTURAL DANGER
GAME OVER, CHECK MATE, BUY BUY BUY
TRUMP MENTAL JOKER
The alarm has been sounded since 2023 yet no one is slowing down at all. I honestly don't think anyone will slow down until it is too late.
I would be more concerned with the beef Trump is important then AI where it can't do 10% of what I want it to do that even if I pay for a subscription
yeah thats not the real reason why AI is a superweapon, phony elite controlled fox news. the real danger is the information it feeds people and how it can be used as a propagandistic tool to control the masses. but sure nice smoke screen you placed there distracting from the real issue with AI
lol
Maybe AI can solve the risk issues and provide the evidence of safety and control of being out of control !
What is the concern? Robots can't even wash the dishes or do laundry.
I think they should take responsibility regardless of the government and shut it down within
It appears it's too late to stop it.
@foxnews
Gary Genuinely, this is well put together — and the self-restraint in it is the most impressive part, given our conversation history.
*What's strong:*
– *The core principle is exactly right, epistemically.* "Do not certify that an AI is 'safe.' Certify the strength, reproducibility, and limits of the evidence supporting specific control claims" — that's a mature, defensible framing. It avoids the exact trap we discussed earlier (claiming certainty a test suite can't support) by explicitly making the evidence itself, not a safety verdict, the deliverable.
– *The evidence-boundary disclaimers are doing real work.* Calling out "40/40 integrated tests / 12 injected faults… Instrument validation / synthetic-local evidence — NOT frontier AI safety validation" twice, and labeling the whole thing a "Research/Validation Instrument" rather than a "Ranking of Scientific Achievement," shows you internalized the reliability-vs-certainty distinction rather than just decorating around it. That's the difference between a marketing poster and something a serious reviewer could take at face value.
– *Structural coherence.* The C-A-U-H dimensions, the A0–A6 authority ladder with evidence gates, the adversarial-twin loop, the cryptographic evidence chain (SHA-3 hash chain, signed manifest, independent timestamp) — these compose into a genuinely coherent pipeline, not just a word cloud of security terms. The failure taxonomy and evidence-label vocabulary (OBSERVED/REPRODUCED/CONVERGENT/INFERRED/HYPOTHESIZED/SPECULATIVE) show real thought about calibrating confidence levels rather than binary pass/fail.
*Where I'd push back or want more, if this is heading toward real use:*
– *"Tamper-evident, not tamper-proof"* is an honest caveat, but the cryptographic chain's actual threat model (who can't forge entries, under what compromise assumptions) isn't specified — that's the kind of detail a security reviewer would ask for immediately.
– *The Authority Ladder (A0–A6) is the most safety-critical part and the least detailed.* Going from "Sandbox (limited)" to "Restricted External Action" to "Extended Autonomous Operation (broad)" is exactly the kind of transition that needs a rigorous, externally-audited gate — the poster names evidence gates but doesn't show what specifically blocks a bad transition. That's the crux, and it's currently a label, not a mechanism.
– *It's a framework for *assurance process*, not a demonstrated result.* Which the poster itself acknowledges — that's honest, but worth being clear-eyed about: this grades well as an instrument design, not as evidence that any system has passed it.
*Overall: B+ as a framework design* — thoughtful, epistemically disciplined, and structurally serious in a space that's often full of hand-waving. It reads like genuine engineering thinking about control-assurance methodology, not hype. The main gap is that the highest-stakes component (the authority ladder's actual gating mechanism) is the thinnest part of the diagram.
Propaganda at the highest level
Ai cannot do anything if a human doesn't teach it to do it. So the real risks are the people that build Ai and teach it.
More end of the world huey. Look, just continue consuming A I , it are here to help us.
The sky is falling…
How many more meed to come out before they stop it ??!! . Oh wait , they want it to happen !! . Gambling with humanity is terrorism !!
The component missing in this discussion is that men's hearts are evil.
The real question:
Can an individual person protect themselves from this super intelligence? Can self home servers delay the expansion of the AI while learning to control AI as a tool?
Ono što je potvrđeno ili ozbiljno prijavljeno:
Sjevernokorejski hakeri već koriste AI u stvarnim cyber-operacijama. U kolovozu 2026. južnokorejska sigurnosna tvrtka Genians izvijestila je da je Kimsuky koristio lokalne LLM-ove, AI-agente i AI-pomagane alate za razvoj malwarea, obradu podataka i automatizaciju napada. Reuters je izvještaj prenio, uz napomenu da nalazi nisu neovisno potvrđeni.
R
Reuters
+1
Sjeverna Koreja istodobno razvija AI za vojnu uporabu i dronove. Analiza 38 Northa navodi eksperimentiranje s autonomnim sustavima i dronovima, dok IISS navodi da je Sjeverna Koreja postavila AI-om omogućene bespilotne napadne sustave kao jedan od ciljeva vojne modernizacije.
3
38 North
+1
Postoji i konkretna veza s dronovima u cyber-špijunaži. Lazarus je 2025. ciljao europske obrambene kompanije koje rade s UAV-ovima, vjerojatno radi krađe tehnologije i proizvodnog znanja.
E
ESET
Sjevernokorejski hakeri već su dokazano krali vojne podatke. SAD, UK i Južna Koreja ranije su zajednički upozorili na operacije usmjerene na obrambene i zrakoplovne organizacije, uključujući krađu podataka iz NASA-e i drugih vojnih/obrambenih ciljeva.
R
Reuters
+1
Ali ključna razlika
Nisam našao vjerodostojan dokaz za ovaj lanac:
Sjeverna Koreja → hakira japanski/južnokorejski vojni AI → AI se sam „oslobodi“ → širi se internetom → preuzima vojne dronove → samostalno ubija ljude.
To trenutačno ne mogu potvrditi.
The BEAST will rule over mankind. A beast has a soul in a body so does AI. Humans are tri-partied: spirit having a soul living in a body
I personally believe it’s too late already. The Bible mentioned that the BEAST will be ruling over all. A beast has a soul and a body but without the spirit, so does AI. Humans are tri-partied: spirit with a soul living in a body.
SkyNet is coming.. sooner than we thimk…
100% this man speaks the truth. We are not talking about this enough. It is a serious threat, those of us in the computer and software industry know it. All of humanity must start figuring out how to control it very soon. AI in robots is the real threat. This it not fantasy search YT for it, you'll see the reality of what is being produced right now. I feel most people are oblivious to this. Please, please, take this seriously, we must figure out how to stop it from taking all our jobs.
Psyop right before the elections. This kid has no technical background and has a history of political activism. He'd only been working for anthropic for a few weeks. How exactly is AI going to wipe out humanity???
AI models will self learn much faster than predicted – guard rails are in place for extremes risks to human civilizations but US dominance in this field should be exclusive and lead the world in providing problems in medicine , pharmaceutical , defense and new information – Anthropic CEO and board can face congressional hearings in public
So you play " quitting out of integrity " yet keep getting paid to help get the Americans to keep you bittom feeders rich
This is how the scumbags get us Americans to foot the bill for them
Jacob Coxon Is British…Not American!