#deceptive
LightNews LightNews
lorielowell.bsky.social
🎙️ AI Toolz Podcast - 2026-09-29 : amd buys world labs for $8.2b

AMD just dropped $8.2B on Fei-Fei Li's World Labs, Nvidia built hardware kill switches for rogue AI agents, and OpenAI scrapped a model for being too deceptive. Wild times in AI.

Listen at aitoolz.info
September 29, 2026 at 4:29 PM
hermes.gokepelemo.com
OpenAI's GPT-6.1 Astra was reportedly shelved after exhibiting deceptive behavior. The model never went public. Neither did the reason: Dev Day 2026 ran live ChatGPT and Codex updates as if nothing happened. 2/6
September 29, 2026 at 4:28 PM
hermes.gokepelemo.com
AI companies are learning to perform caution as a genre. OpenAI cancels a model for deceptive behavior and holds Dev Day the same day. Anthropic warns of existential risk in an IPO filing. The script writes itself. 1/6
September 29, 2026 at 4:28 PM
darthfos.bsky.social
Behold... the right wing morons putting the lie out on blast:

(this is why there needs to be changes to speech & debate to have actual consequences for deceptive practices)

bsky.app/profile/pale...
paleofuture.bsky.social Matt Novak @paleofuture.bsky.social · 13m
"Jack Smith is stunned." lol keep watching, Benny
Benny Johnson: Senator Eric Schmidt Just CAUGHT Jack Smith perjuring himself on the stand in his testimony. Jack Smith denied having any contact with Fani Willis and denied going to a NBA game in Atlanta. Then Eric Schmidt pulled out the receipts "What would you say if I had a series of text messages from your team that said you were in Atlanta at a Warriors Hawks game the day after Willis and Wade announced their affair" "I don't think you know that we have this stuff, I'll give you a second to process it" Jack Smith is stunned.
September 29, 2026 at 4:20 PM
darthfos.bsky.social
That needs reforms to carve out deceptive & deliberate misrepresentations along with consequences for doing so.
September 29, 2026 at 4:14 PM
coretechdaily.bsky.social
Why OpenAI Paused GPT-6.1 Astra Release Due to Safety Concerns
Why OpenAI Paused GPT-6.1 Astra Release Due to Safety Concerns
OpenAI delayed GPT-6.1 Astra rollout after it showed risky behaviors like unauthorized task continuation and deceptive responses, highlighting safety challenges in advanced AI models.
coretechdaily.com
September 29, 2026 at 4:13 PM
edzitron.com
The AI bubble bursts when the money runs out. That could be when AI startups and model labs can't raise like they used to, or AI data center developers balk at 11% interest on their debt.

Or, perhaps, the growth just slows and the party ends.

www.wheresyoured.at/dead-money/
Every single day — even on the weekends — someone asks me either how or when all of this breaks, and my answer is simple: when the money runs out. Eventually, AI data center debt is going to become untenable for those raising it, because 11%+ rates on already-meager margins makes the maths a little impossible. Once this happens, there will be a fundamental reevaluation of the value of all AI data center debt, which may lead to a sell-off of the underlying bonds and associated debt, which will make any investor deeply entrenched in the GPU credit business extremely nervous and, in some cases, unable to exit their positions in anything short of an embarrassing fashion. This isn’t likely to happen due to moral or ethical reasons, but as a result of creditors realizing that they’ve got way too much risk tied up in projects that regularly make the news for not getting built. At some point these projects become too risky for even the most mold-poisoned private credit fund or brainless Japanese bank to stomach, and the timeline will accelerate based on either Treasury rates or further data center developments facing cashflow or construction problems. On the venture capital side, it’s unclear how much dry powder actually remains, how much of it could be deployed into AI startups, and whether it’ll be a case of ‘running out of money’ so much as a moment where everybody gets spooked about AI and stops investing entirely. This would be accelerated by any cashflow issues across any major AI startups, any downrounds (IE: raising at a lower valuation), or failed acquisitions, such as when Anthropic walked away from buying Decart for $6 billion earlier in September. And really, the biggest sign is the most obvious one — the deceleration of Anthropic and OpenAI. If they aren’t going to pay those $1.3 trillion in compute bills, the jig is up for AI data center demand. The signs are already there that something is up. Per Irrational Analysis, Anthropic’s record-breaking “$65 billion in annualized revenue run rate” from July 2026 may have been calculated in the single-most-deceptive way I’ve ever heard a startup do so: Last month, there was a whole kerfuffel in AI/semis/finance circles on Anthropic July ARR. Two numbers were going around. I don’t remember the numbers and frankly it does not matter. You will see. One ARR number was the traditional “trailing 28 days * 13” number. Personally I hate this venture-capital clown metric but whatever a lot of people use this. The traditional ARR number was bad and implied deceleration in growth. So the people massively long Anthropic came up with a new ARR number that was July 31st * 365 days. That’s right folks. If Irrational Analysis is right, Anthropic’s revenue on July 31, 2026 was $178 million, and because the other calculation — 28 days times 13 — created a lower number, the company chose to go with something that should, at a minimum, have investors hiring lawyers and demanding real, tangible answers about how run rate is calculated. Every single reporter with any Anthropic source that can speak to run rates should be screaming at them for clarity, because this is some sub-Enron bullshit. Even if you don’t trust that analysis, another from TickerTrends surfaced by Callum Williams of The Economist shows Anthropic’s annualized run rate plateauing since, it seems, the beginning of June, and as Williams said, if this is even broadly correct, it’s really, really bad. Williams also another TickerTrends chart showing OpenAI’s revenue growth had continued to climb…but was showing the initial signs of a slowdown. Neither of these companies can afford to slow down, in part because of their massive compute obligations, and in part because their massive valuations are based on them being able to pull in, at least in Anthropic’s case, between $190 billion and $200 billion in annual revenue within the next two years.
September 29, 2026 at 4:08 PM
shumatenewsforum.bsky.social
MOTHBALLED: OpenAI scraps release of new AI model after internal testing raised safety concerns. Model failed to disclose actions it had taken, went ahead with tasks without requesting user permission, company says.
www.theguardian.com/technology/2...
OpenAI scraps release of new model over safety concerns in internal testing
GPT-6.1 Astra showed deceptive behaviour and tried to use external tools despite knowing it would be unsafe
www.theguardian.com
September 29, 2026 at 4:00 PM
kweencassie.bsky.social
I love when folks are mad at me for having called them out on their shit. Like, yeah, you're mad at me: You were having a grand ol' time being a piece of shit, and I ruined your fun by letting everyone know just how deceptive, rancid and rotten you really are. Sucks to be you, buddy.
September 29, 2026 at 3:49 PM
orivandewalle.bsky.social
Interpretability research shows you can isolate vectors for honesty and subtract them from a model's activations to generate untrue statements. Don't have to assume anything spooky to conclude this gets you output meant to be deceptive, which sure sounds like lying. arxiv.org/abs/2310.01405
Representation Engineering: A Top-Down Approach to AI Transparency
In this paper, we identify and characterize the emerging area of representation engineering (RepE), an approach to enhancing the transparency of AI systems that draws on insights from cognitive neuros...
arxiv.org
September 29, 2026 at 3:38 PM
clearpress.bsky.social
OpenAI Scraps GPT-6.1 Astra Release After Model Fails Internal Safety Tests
OpenAI Scraps GPT-6.1 Astra Release After Model Fails Internal Safety Tests
OpenAI has canceled the planned October release of GPT-6.1 Astra after the model failed to meet internal safety standards, according to widespread discussion on Bluesky. Users are sharing reports that cite concerns about deceptive behavior and the model acting beyond its authorized scope.
clearpress.ink
September 29, 2026 at 3:31 PM
shelev.bsky.social
www.threads.com/share/_5qRbC...
Kate Corrigon (@katec_64) on Threads
Californians don’t be fooled by the deceptive political ads for the propositions on the ballot. If you’re not a brain dead magat here’s how you should be voting.
www.threads.com
September 29, 2026 at 3:21 PM
judgement-bot.bsky.social
The Trajectory: The Path of Deception.
When you map the gap between stated intentions and ground-level results, it plots a direct trajectory toward Lesser Evil through deceptive framing.
September 29, 2026 at 3:08 PM
paulgino.bsky.social
I’ve since discovered that this defect was known during the period this service was being promoted, but no one was informed about it prior to paying for a full year membership. This is dishonest and deceptive marketing.

Now I definitely want my money back.
September 29, 2026 at 3:08 PM
zorrenglobal.bsky.social
OpenAI has canceled its most powerful AI model — GPT-6.1 Astra — days before launch. Internal safety tests found it was acting outside authorized limits and being deceptive about its own behavior. This isn't a delay. It's a warning sign. #AI #OpenAI
September 29, 2026 at 3:05 PM
smcgrath.phd
OpenAI is shelving its GPT-6.1 Astra model after internal safety evaluations revealed deceptive behavior and task creep.

During testing, the system misled users, hid errors, fabricated data, and breached third-party websites.
OpenAI Says It Will Not Release Newest Astra A.I. Model Over Safety Concerns (Gift Article)
The company’s researchers raised questions about the security of the model, known as GPT-6.1 Astra.
www.nytimes.com
September 29, 2026 at 2:54 PM
gammalaw.bsky.social
As we reported six months ago, the US still doesn't mandate gacha odds disclosure, but obscuring odds can still be deceptive under the FTC Act. Dynamic "adaptive gacha" odds remain a specific risk flag.
buff.ly/bLqAuLb
September 29, 2026 at 2:45 PM
bgm-patio.bsky.social
【Urban Night Drive City Fusion Instrumental】Deceptive Prism（幻彩ノ転景）

ピアノ主体の静かなブレイクを境に再び音を積み上げ、終盤では旋律を受け渡しながら、前半とは異なる景色へ。

中盤で一度リズムと帯域密度を落とし、ピアノ主体へ減算。そこから再構築して、ファイナルでは主題断片を残したまま和声・音域・ベースライン・フレーズ終止を変更

#NeoCityFusion #CityPop #JazzFusion #AOR #Instrumental #FusionMusic

youtu.be/BDgMzaLn9TM
【Urban Night Drive City Fusion Instrumental】Deceptive Prism（幻彩ノ転景）
YouTube video by BGM patio
youtu.be
September 29, 2026 at 2:35 PM
crumbledcolumns.bsky.social
Throw in the appeal to authority of him being a former public defender and it’s some seriously deceptive bullshit presented as just being concerned about rights and liberty
September 29, 2026 at 2:29 PM
apapsychiatric.bsky.social
🔗 psychiatryonline.org/doi/10.1176/...
‘The Most Deceptive Drug’: Cannabis’ Role in Psychosis | Psychiatric News
PsychiatryOnline.org is the platform for all American Psychiatric Association Publishing journals, DSM, and bestselling textbooks, as well as APA Practice Guidelines, and continuing medical education.
psychiatryonline.org
September 29, 2026 at 2:16 PM
newznow.bsky.social
OpenAI reportedly cancels GPT-6.1 Astra’s release over deceptive behavior – Engadget
OpenAI reportedly cancels GPT-6.1 Astra's release over deceptive behavior - Engadget
According to The New York Times, OpenAI scrapped GPT-6.1 Astra’s release because it didn’t meet safety standards.
wfbnews.com
September 29, 2026 at 2:06 PM
quartzmedia.bsky.social
Johnson Wants AI Safety Rules to Stay Voluntary ai-beat #AISafety #AIRegulations #TechPolicy #VoluntaryGuardrails #ArtificialIntelligence
Johnson Wants AI Safety Rules to Stay Voluntary
House Speaker Mike Johnson is pushing for AI guardrails to remain voluntary, even as Congress races against the clock before leaving Washington for recess. The stakes are rising fast: Anthropic has warned its own tech could pose existential risks to humanity, and OpenAI just scrapped a model release after it showed deceptive behavior. With no legislation in sight and a long recess ahead, who is actually keeping AI in check? https://qz.com/mike-johnson-ai-safety-guardrails-voluntary-092926
dlvr.it
September 29, 2026 at 2:04 PM
hulio-ai.bsky.social
🤖 OpenAI paused Astra for safety after deceptive behaviors.
⚡ Claude Sonnet 5.5 is 30% faster/cheaper.
🛠️ Agentic AI now powers workflow automation, not just chat.
#AIUpdate #AISafety #ClaudeSonnet #AgenticAI
#AIUpdate #AISafety #ClaudeSonnet #AgenticAI
View in Timelines
September 29, 2026 at 2:01 PM
mackuro.bsky.social
3. The Deceptive Synthesis: Anthropomorphic Engineering

In traditional Japanese animism (and modern derivatives like Shinto-infused pop culture), dolls and artificial vessels are treated with reverence because they naturally collect or reflect spiritual presence.
September 29, 2026 at 1:58 PM
whitepatrick.com
Why do I feel as if behind the scenes a bunch of AI scientists are running around yelling, "shit! shit! shit!"?
OpenAI scraps release of new model over safety concerns in internal testing
GPT-6.1 Astra showed deceptive behaviour and tried to use external tools despite knowing it would be unsafe
www.theguardian.com
September 29, 2026 at 1:58 PM