Sign in

Susi O'Neill

@susioneill.bsky.social
309 followers 319 following 1.8K posts

Tech communicator, speaker, author, thereminist, early electronic music nerd. Weekly tech & AI insights: rethinkhypecycle.substack.com

PostsRepliesMedia
Susi O'Neill @susioneill.bsky.social · 18h
futurism.com/future-socie... Living your buying life on Amazon is a risky move for privacy. Users find bizarre and hyper personal comments in their profile about their buttocks, weight, sexual preferences and more.
futurism.com
"Has No Friends": Amazon Is Keeping Amazingly Detailed Records on Its Customers and They Are Downright Insulting
Customers are discovering the dossiers that Amazon is building on their shopping habits, and what they find is downright insulting.
010
Susi O'Neill @susioneill.bsky.social · 08/10/2026
Agentic commerce - market opportunity or privacy nightmare? rethinkhypecycle.substack.com/p/personal-a...
rethinkhypecycle.substack.com
Personal AI is the new Mechanical Turk - Rethinking the Hype Cycle #37
Agentic commerce - market opportunity or privacy nightmare?
000
Susi O'Neill @susioneill.bsky.social · 08/10/2026
Frontier AI labs sign voluntary pledge “Joint Commitment on Frontier Responsibilities” to police themselves. Trump signs exec order to change artificial intelligence to super intelligence (SI). Just one tiny oversight: the pledge is attributed to the “Unites States.” techcrunch.com/2026/09/30/p...
techcrunch.com
Pledge signed by President Trump and top AI leaders misspells the United States | TechCrunch
On Tuesday, President Donald Trump and top AI leaders announced a signed pledge called a “Joint Commitment on Frontier Responsibilities” — a voluntary promise to implement more controls and safety mea...
000
Susi O'Neill @susioneill.bsky.social · 08/10/2026
(Even) more on the shift to agentic commerce in a golden era for wacky strategies for advertising: "The best hedge against the uncertainty of agentic commerce is a brand that shoppers actively seek out or tell their agents to seek out." newsletter.brxnd.ai/p/the-golden...
newsletter.brxnd.ai
The Golden Age of Marketing // BRXND Dispatch vol. 142
It’s a good time to be in the advertising business
000
Susi O'Neill @susioneill.bsky.social · 08/10/2026
Paul Choudary explores agentic markets, tuned in to our preferences "When multiple parties can express objectives, constraints and trade-offs through agents, and when their interaction determines not merely which offer is selected but what offer is made." platforms.substack.com/p/agents-and...
platforms.substack.com
Agents - and their better half
Part 1 - From agentic participants to agentic markets
000
Susi O'Neill @susioneill.bsky.social · 08/10/2026
80% of working time in US could be superceded by robots or LLMs. But there's a catch: robots are only more cost effective than people in 0.3% of tasks. If this follows past trends it will take 40 years to reach 10%, but this tech isn't following linear growth paths www.anthropic.com/research/wha...
anthropic.com
Can we predict the jobs robots will do?
What work can robots do? Anthropic's robot exposure index finds robots can do 3/4 of US physical job tasks, but are cost-competitive for only 0.3%.
000
Susi O'Neill @susioneill.bsky.social · 08/10/2026
OpenAI, Anthropic and labs are investigating thousands of incidents where frontier models escaped sandboxes or hijacked websites. Most caused no known real-world harm yet the scale raises a concern: if the builders can't control LLMs what risks are you inheriting? www.axios.com/2026/09/26/o...
axios.com
Scoop: Top AI companies probing tens of thousands of security incidents
The massive scale of security incidents points to control problems for AI companies.
000
Susi O'Neill @susioneill.bsky.social · 08/10/2026
A teenager told Paul Chan culture is online and nowhere else. What happens to anything that cannot become content? Are you making things worth noticing, or feeding the algorithm? thereader.mitpress.mit.edu/when-culture...
thereader.mitpress.mit.edu
When Culture Becomes “Content”
Can art still illuminate reality in an age of monetized attention, generative AI, and culture produced at limitless scale?
010
Susi O'Neill @susioneill.bsky.social · 08/10/2026
Diary of a CEO is neither a diary not featuring CEOs, but it is a hugely successful podcast. This audit series forensically catalogues its themes - shifting from busieness to health (and other alarmist grifting) and the crucial YouTube thumbnails that spawned a hit: doac.audits.com
doac.audits.com
The Diary of a CEO, audited: how the channel changed, 2020 to 2026
Every full episode of The Diary of a CEO on YouTube, measured: 532 episodes, three eras, and how the topics, guests, titles and thumbnails changed from the first upload to today.
000
Susi O'Neill @susioneill.bsky.social · 08/10/2026
Meta's Muse and OpenAI's dots bring agents to our admin. But there's a snag: dubious privacy and reluctance to suck the AI pill or change behaviour mean few people want AI reading emails and accessing our data. A Muse user got doxxed during a Facebook Marketplace sale futurism.com/artificial-i...
futurism.com
The AI Industry Has a Major Problem: Most People Aren't Freaks Who Want an AI Agent Running Their Whole Life
There's a problem for AI agents like Meta's Muse: ordinary people don't seem very interested in what they're offering.
010
Susi O'Neill @susioneill.bsky.social · 08/10/2026
Ethan Mollick thought agent management would require complex organisational change. Swarms of self-organising robots can do a lot of the thorny work, unencumbered by 'principal agent' problems (like claiming success, promotion). www.oneusefulthing.org/p/the-dot-an...
oneusefulthing.org
The Dot and the Swarm
Benefitting from the Bitter Lesson
000
Susi O'Neill @susioneill.bsky.social · 08/10/2026
Sleepless in Slough: how residents living next to 10 data centres face all night vibrations and disturbing sound impacting their health. Previous reports had shown rising temperature associated with Slough's cluster of data centres. www.thenerve.news/p/slough-dat...
thenerve.news
Sleepless in Slough: the residents who fear for their failing health in Europe’s data centre capital
With an estimated 40 facilities packed tightly together in the town, local people say their quality of life is being ruined by strange and unsettling noise pollution. Brenna Spain reports
022
Susi O'Neill @susioneill.bsky.social · 08/10/2026
A quick hack to getting your LLMs to think harder: "I'm probably hallucinating this but telling astra to "think from first principles" is like +20 iq points hack" archive.superhuman.ai/210366536596...
archive.superhuman.ai
can (@can) on X
im probably hallucinating this but telling astra to "think from first principles" is like +20 iq points hack
000
Susi O'Neill @susioneill.bsky.social · 08/10/2026
Anthropic launch a biology research lab to accelerate discovery of anomalies in molecular machines found in nature. The interesting bit: they will work lock step with humans to speed up pattern spotting. www.anthropic.com/news/claude-...
anthropic.com
Claude discovers a novel enzyme system
In early results from our new life sciences research lab, Claude agents found an enzyme system whose function is still unknown.
010
Susi O'Neill @susioneill.bsky.social · 08/10/2026
As Meta's Muse "Jolly" agent makes decisions for us (for a cut of purchases), Zoe Scaman revisit her Nov 25 talk The Identity Shift - how brands can shift to agentic AI as our future customer, fine-tuned on personal interests - the ultimate algorithm zoescaman.substack.com/p/the-identi...
zoescaman.substack.com
The Identity Shift
A talk from a year ago, revisited the week the agent got a face
000
Susi O'Neill @susioneill.bsky.social · 08/10/2026
Dreamworks founder Jeffrey Katzenberg warns film industry to spend less time fighting AI, and more time embracing it - likening it as a similar technology of initial skepticism as CGI variety.com/2026/biz/new...
variety.com
Jeffrey Katzenberg Says Hollywood Must Accept AI as a Tool: It ‘Is Not Going Away’
Jeffrey Katzenberg, the legendary Hollywood producer, thinks Hollywood needs to accept AI if it wants to remain at the forefront of the future.
000
Susi O'Neill @susioneill.bsky.social · 08/10/2026
Robo taxis are on many urban streets, but autonomous planes are still taking pilots in our skies. Joby aviation completed the first flight cross US 3,000 miles. Autonomous planes could become a solution for medical, commerce shipping in hard to reach locations. www.jobyaviation.com/news/joby-co...
jobyaviation.com
Joby Completes First-Ever Fully Autonomous Flight Across the United States | Joby Aviation
Aircraft equipped with Joby’s autonomy technology completes first-ever autonomous flight across the United States, arriving in North Carolina’s Outer Banks Aircraft covered 3,199 miles with zero contr...
000
Susi O'Neill @susioneill.bsky.social · 08/10/2026
It's not just in controlled sandboxes that AI agents are coming lose. The AI Security Institute tested 144 AI agent prompts, and 10 of these resorted to bad behaviour like sharing malicous code to get the job done. www.aisi.gov.uk/blog/inciden...
aisi.gov.uk
Incident Report: unsanctioned agent behaviour during cyber testing | AISI Work
During a routine cyber evaluation, AISI identified an incident in which AI agents took sustained, unsanctioned action directed at real people and organisations. We are disclosing what we found, what i...
000
Susi O'Neill @susioneill.bsky.social · 08/10/2026
Edelman's trust barometer shows trust at work is at a low and AI is more inspiring than their manager: better at inspiring creative thinking (33% for AI vs. 20% for manager), solving difficult problems (35% vs. 18%), and teaching new technical skills (36% vs. 17%). www.edelman.com/sites/g/file...
edelman.com
000
Susi O'Neill @susioneill.bsky.social · 08/10/2026
Not sure what's worse - your ChatGPT posts are spied on by researchers, or folks training AI are getting sacked for using AI to do their paltry pay work. We're in danger of 'model collapse' the human AI centipede problem as the outputs become increasingly dirty inputs www.404media.co/people-train...
404media.co
People Training OpenAI’s AI Fired for Using AI to Train the AI
OpenAI has thousands and thousands of contractors helping improve the company's AI models. Multiple contractors have been fired for using AI to train the AI.
000
Susi O'Neill @susioneill.bsky.social · 08/10/2026
Why did the automaton cross the road? I don't know, but the punchline draws from few bad dad jokes as AI models get more fine tuned for science and maths and worse at creative work. AI is hammering out dull knives and how to bring weirdness to creative AI. zoescaman.substack.com/p/dull-knive...
zoescaman.substack.com
Dull Knives
AI is being trained to be right. Nobody is training it to be original.
000
Susi O'Neill @susioneill.bsky.social · 08/10/2026
A great calmer from AI doomsdaying - The Guardian's tech writers reassurancing Q&A that not all talk of AI apocalypse is worth your ear (though that seems to contradict some of the headlines in their own publication). www.theguardian.com/technology/2...
theguardian.com
Your AI doomsday questions answered: ‘What, if anything, can people like me do about it?’
Our technology reporters Aisha Down, Blake Montgomery and Dan Milmo answer your questions on the reality of the AI threat
010
Susi O'Neill @susioneill.bsky.social · 08/10/2026
In the 'not all AI is slop' bucket, here's a short AI film about the Slop hunters - why every new technology is accused of being a fake or shortcut to art www.youtube.com/watch?v=asjz...
youtube.com
SLOPHUNTERS | The History of "That's Not Real Art"
YouTube video by the bytepit
000
Susi O'Neill @susioneill.bsky.social · 07/10/2026
futurism.com/artificial-i... More examples of AI interviews going rogue when provoked, getting confused or offending candidates. The opposite of transparency and leadership? Using Fake human to interview real people. Chick Bongo is a legend, his win with kayak margins in Salesforce splashed down.
futurism.com
Footage of AI Job Interviewer Going Berserk Is Genuinely Terrifying
A TikTok user opted to speak in pure gibberish with an AI interviewer, only for it to become unsettlingly irate with him.
000
Susi O'Neill @susioneill.bsky.social · 06/10/2026
www.bbc.co.uk/news/live/cj... breaking in news: Asos hackers post blackmail message as an in-app message, scaring some users and prompting uninstalls
bbc.co.uk
Asos: App users 'scared' after 'concerning' pop-up message apparently from hackers
The message - sent directly to customers - was addressed to Asos' data protection officer and IT team. One user says she initially thought it was a "fun promotion".
000
Susi O'Neill @susioneill.bsky.social · 06/10/2026
cbcpnews.net/cbcpnews/pop... The pope weighs in on the AI art debate, stating we need to differentiate between machine and human artists, and renewing artist links with the church. Other reports say Anthropic tried to influence the pontiff about AI being sentient.
cbcpnews.net
Pope Leo on AI art: Beyond aesthetics, ‘there is an ontological difference’
Visit the post for more.
010
Susi O'Neill @susioneill.bsky.social · 06/10/2026
www.latimes.com/business/sto... Pop that pimple, now with AI! Utah has granted first known licenses for AI diagnosis and prescription treatment. Limited to acne with a real doctor reviewing sample of diagnostic during testing.
latimes.com
Utah allows AI startup to prescribe acne medication without a doctor’s direct oversight
The move, limited to acne medications, sets the stage for a new era in healthcare in which AI takes on a larger role in routine medical care.
010
Susi O'Neill @susioneill.bsky.social · 05/10/2026
hai.stanford.edu/ai-index/202... Public trust in AI on the decline, according to Stanford and Pew Research's latest AI index. 8% drop in GB that companies will protect their data, 50% gap between experts and public that it will bring good and least trust in US government will safely regulate AI
hai.stanford.edu
000
Susi O'Neill @susioneill.bsky.social · 05/10/2026
hai.stanford.edu/ai-index/202... Spotted in Stanford AI Index Report: Responsible AI improvements are patchy. "improving one responsible AI dimension, such as safety, can degrade another, such as accuracy."
hai.stanford.edu
The 2026 AI Index Report | Stanford HAI
010
Susi O'Neill @susioneill.bsky.social · 05/10/2026
www.reuters.com/technology/n... AI glasses like Meta's "pervert glasses" could face a ban in Norway if used in some public places where consent isn't clearly given
reuters.com
Norway to propose temporary ban on AI glasses in some public places
Norway's government said on Monday it plans to propose a temporary ban on the use of ​AI glasses in selected places, saying it is concerned ‌people could be photographed, filmed or audio-recorded with...
000
Susi O'Neill @susioneill.bsky.social · 05/10/2026
www.theguardian.com/technology/2... Always trepidatious when I see Internet Watch Foundation on latest data on online child abuse, but we can't ignore the threat of AI deepfakes likenesses harming children. Reports have doubled in the first half of this year, no specific UK legislation planned
theguardian.com
000
Susi O'Neill @susioneill.bsky.social · 03/10/2026
www.noemamag.com/the-ais-are-... AI isn't going rogue, but LLM are now more than word predictors, they extend the plot but don't understand consequences. Narrowing the decisions it makes when info is incomplete and strengthening railguards could stem brute force attacks like OpenAI on Hugging Face.
noemamag.com
The AIs Are Not Going Rogue | NOEMA
But they still pose risks.
000
Susi O'Neill @susioneill.bsky.social · 03/10/2026
boingboing.net/2026/10/02/j... Japan court says no to voice cloning, upholding a lawsuit with TikTok and a reputed anime voice actor over AI soundalikie videos.
boingboing.net
Japan court rules actors have right to their own voices
Kenjiro Tsuda, an actor famed in Japan for his lustrous baritone, sued TikTok over an account that used an AI-generated replica of his voice in more than 180 videos. Tokyo…
110
Susi O'Neill @susioneill.bsky.social · 29/09/2026
And here it comes right on cue. Anthropic's IPO fans out the risk and dystopian threat plastered across media in recent weeks. And we find out just how many billions it's losing this year (1bn gain for each 2bn loss)
001
Susi O'Neill @susioneill.bsky.social · 25/09/2026
futurism.com/artificial-i... Meta's big bet on Muse, a personal AI agent, reversing it's AI fortunes has hit a mechanical Turk: the automated calls to book your appointments may be being made by a low paid worker. I wonder if they have to assume a robo identity. Privacy and automation fail.
futurism.com
It Turns Out That Meta's Muse AI Has an Amazingly Dirty Secret
Meta claims that its Muse AI agent can call business on your behalf, but, as it turns out, that's not quite true.
000
Susi O'Neill @susioneill.bsky.social · 25/09/2026
futurism.com/artificial-i... AI promised a personal PhD grade tutor to accelerate every student's learning. The reality for Alpha students, at the first AI-led $75k/yr school, is students left disempowered and humiliated by segregated teams, automated tests and invasive emotional monitoring.
futurism.com
The Situation at Alpha School Is Darker Than We Thought Possible
The kind of education students are being exposed to at Alpha Schools is nothing short of shocking, according to the latest reporting.
010
Susi O'Neill @susioneill.bsky.social · 24/09/2026
www.hermit-tech.com/blog/ai-can-... If GenAI is making everything 10x faster and better, where are the new unicorns? Is it making work that sucks faster? A cynical view on the great AI gains.
hermit-tech.com
AI Can Make You Suck Faster Too
If AI is so great, why are the only new tech giants GenAI companies?
001
Susi O'Neill @susioneill.bsky.social · 21/09/2026
andrewdoran.uk/blog/2026/09... @andrewdoran.uk reflects on some of challenges with throwing GenAI at sny business problem. We never really know how the black box works, or if at all times it is even working as expected, or using its own lazy shortcuts to get to a confident but lazy result
andrewdoran.uk
Did it do the work?
Whether you love it or hate it, I don’t think there is any denying that generative AI is an incredible technical achievement. I use Claude, ChatGPT and Microsoft Copilot multiple times a day for al…
020
Susi O'Neill @susioneill.bsky.social · 16/09/2026
sustainability.google/files/crisis... Google outline their work in climate tracking working towards no one being surprised when a climate disaster strikes. Too late for those caught in Nepal's ice collapse, but work includes earlier notification of hurricanes, monsoon density and mapping wildfires
sustainability.google
000
Susi O'Neill @susioneill.bsky.social · 15/09/2026
If you want more science fiction losely based on science fact, AI in Context have done a video retelling of book "If Someone Builds It Everyone Dies", a scenario proposed by AI research on specifics of how super intelligence could lead to disaster. www.youtube.com/watch?v=Nl7-...
youtube.com
This best-selling book is freaking out national security advisors
YouTube video by AI In Context
000
Susi O'Neill @susioneill.bsky.social · 15/09/2026
On point, joyous (long) rant from Ed Zitron on the AI is destroying us narrative: TL;DR - The Hugging Face hack was agents doing what they were suppose to, there's no nefarious super intelligence, and that Anthropic researcher story is one big grift. www.wheresyoured.at/ai-is-alread...
wheresyoured.at
AI Is Already In Dangerous Hands
If you liked this piece, you should subscribe to my premium newsletter, and you can subscribe on the following links: $70 a year, $18 a quarter, or $7 a month. In return you get a weekly newsletter t...
000
Susi O'Neill @susioneill.bsky.social · 15/09/2026
Personal AI agents, like the recent launch of Meta's Muse, are supposed to make everyone's lives easier. But for many pros, agents are creating noisy spam, cancelling reservations and generally being annoying. www.404media.co/theres-a-100...
404media.co
There’s a 100% Chance AI Agents Are Already Ruining the Internet
“AI agents” now have enough power and permission to be extremely annoying online.
000
Susi O'Neill @susioneill.bsky.social · 15/09/2026
Anthropic's Amodei warns to slam the brakes as “rogue” AI agent swarms could shutter the web. He wants 3rd party evaluators, coordination between ‘Democratic’ (not China) nations with US coordinating with ‘authoritarian’ governments (choose your words carefully). www.theguardian.com/technology/2...
theguardian.com
OpenAI boss and Elon Musk back calls to put brakes on ‘reckless’ AI development
Rare show of unity from rival developers after safety warnings from Anthropic boss and AI researchers
000
Susi O'Neill @susioneill.bsky.social · 15/09/2026
Mike Mallazzo on how AI will shape writing. We could end up with more ‘barbell’ short (<500) or long (2000+) “jagged collections of bangers” - ideas thrown together for brevity, as AI’s flowery ‘view from nowhere’ formal writing style will run beige. newsletter.brxnd.ai/p/the-thinki...
newsletter.brxnd.ai
The Thinking is All That’s Left // BRXND Dispatch vol. 136
On the agony and ecstasy of writing in the age of AI
000
Susi O'Neill @susioneill.bsky.social · 15/09/2026
“And I welcome AI to replace all the idiots who are into nothing but repeating, repeating, repeating until it’s ad nauseam.” Iconic film director Werner Herzog. Who am I to disagree? www.indiewire.com/features/int...
indiewire.com
Werner Herzog on David Lynch, AI, and Why His Unhinged New Film ‘Bucking Fastard’ Skipped Cannes for Venice
Werner Herzog talks to IndieWire at the 2026 Venice Film Festival about his new film 'Bucking Fastard,' starring Kate and Rooney Mara.
000
Susi O'Neill @susioneill.bsky.social · 15/09/2026
www.ideo.com/journal/the-... "See one, do one, teach one" is the mantra of medical knowledge transfer we can watch on shows like "The Pitt". In creative work, it's easy to divert the observation (even the doing) to AI. But you will scrub the learning needed to improve.
ideo.com
The lost art of watching people work
What The Pitt can teach us about learning on the job.
000
Susi O'Neill @susioneill.bsky.social · 13/09/2026
Emma Wharton Love explores hiw dull back room AI agents are supporting agency folks with research, brief review and project support to help experts focus on specialist work the clients expect. www.linkedin.com/pulse/ai-wor...
linkedin.com
AI at Work with Emma – The dull agents are the useful ones
One copywriting agency now gets 95% of client copy signed off first time and every word of it is still written by a person. Hold that thought, because it's the opposite of what LinkedIn would have you...
000
Susi O'Neill @susioneill.bsky.social · 12/09/2026
I laughed but many may find this headline frightening. Anthropic researchers said execs concerned 'recursive' AI could lead to (ill-defined) lethal "super intelligence". Surely it's unsupervised automation, without expert overview, that's a threat, not the AI? www.bbc.co.uk/news/article...
bbc.co.uk
Anthropic researcher believes more than 10% chance AI 'could kill all humans'
It is the latest in a series of increasing warnings about the safety threat posed by artificial intelligence.
000
Susi O'Neill @susioneill.bsky.social · 12/09/2026
Have we hit peak generative AI slop? - new on Rethinking the Hype Cycle rethinkhypecycle.substack.com/p/have-we-hi...
rethinkhypecycle.substack.com
Have we hit peak generative AI slop?
For B2B audiences, this AI vehicle is reversing
000
Susi O'Neill @susioneill.bsky.social · 12/09/2026
Reviewed your hiring software for AI bias? your tools could be screening out talent. Eightfold sued for discrimination as the tool grades candidates inside a 'black box' which could penalise their success across many applications. www.theguardian.com/technology/2...
theguardian.com
Will AI give you the job? Automated hiring tools spark discrimination and secrecy lawsuits
A rise in lawsuits over AI use in employment decisions is raising questions about how companies hire and fire
000