Topic
LightNews LightNews
Technology US

AI expansion could trigger crisis

3m

Experts and commentators warned that AI’s rapid expansion risked triggering a broad crisis—eroding public trust, enabling deepfakes that could threaten critical systems and prompting calls for government intervention.

by Robert B. Reich

38%
rbreich.bsky.social
The AI bubble keeps getting pumped bigger and bigger while the rest of the economy is gasping for air.
Brace Yourself for the AI Bubble
Robert Reich
www.youtube.com
December 28, 2025 at 11:00 PM

Reposted by Joe R. Feagin

17%
brucedlittle.bsky.social
Bernie Sanders is calling for a temporary pause on AI data center construction nationwide.

He says a pause would "give democracy a chance to catch up" with the rapid build-out.

He's concerned about potential AI-driven job losses and the tech's effect on human relationships.
Bernie Sanders wants to temporarily halt AI data center construction nationwide
"This moratorium will give democracy a chance to catch up with the transformative changes that we are witnessing," Sanders said.
www.yahoo.com
December 29, 2025 at 3:28 AM

by Ruha Benjamin Reposted by Hilary J. Allen

9%
ruha9.bsky.social
“Accepting this story of AI’s inevitability — or that of any technology — is a grave mistake for educators. Not only does it cede tremendous power to tech companies that are unaccountable to teachers, students, or school communities…” 1/2

kappanonline.org/teach-like-a...
Teach like a Luddite - Kappan Online
Embracing new technologies that don’t advance teaching and learning is a mistake. Educators must ask questions — and resist when necessary.
kappanonline.org
December 29, 2025 at 6:40 PM

by Anima Anandkumar

38%

by Bruce Schneier

55%

Reposted by Juan Cole

13%
peark.es
A quarter of firms *think* that AI will lead them to hire less, but 92% say that's not the case with implementation so far.
December 29, 2025 at 3:46 PM

by Hilary J. Allen

12%
profhilaryallen.bsky.social
This article depresses me. There's so much opportunity for pols to capitalize on justified skepticism re the tech industry, but a16z et al learned just how easy it was to buy votes for crypto legislation so they're repeating the playbook for AI & it'll prob work.

fintechdystopia.com/chapters/cha...
December 29, 2025 at 5:58 PM

Reposted by Mark Rice, Alan Richardson

4%
jakecole.bsky.social
There’s a funny Politico story showing that the latest polling on AI shows 80-20 in favor of heavy regulations, an absolute slam dunk platform, but Dems are worried about not winning back mask-off tech CEO donors
www.politico.com/news/magazin...
December 29, 2025 at 1:55 PM

by Pedro Domingos

57%

Reposted by Greg Linden

5%
edzitron.com
The reason we haven’t seen the appropriate level of fear is because, on paper, things are great. These AI companies have sky-high valuations on paper, making VCs look like geniuses...except nobody is buying AI companies. Look at Groq, or Inflection, or Windsurf.
www.wheresyoured.at/the-enshitti...
You see, right now AI startups are big, exciting news for the limited partners funding LLM firms. Things feel exciting because the value of the assets under management (AUM) are going up, which is nothing dodgy, but just how VCs value things and if they are valuing AI stocks, that is how their fees are paid. Investing early in OpenAI allows a VC — or even an asset manager like Blackstone, which invested in 2024 — to say it has a big holding and a big increase in its AUM. We are currently in the sowing stage. Nevertheless, AI stocks make VCs who bet on them two years ago look like geniuses on paper. You got in early on OpenAI, Anthropic, Cursor, Cognition, Perplexity or any other company that loves to burn several dollars per dollar of revenue, you have a big, beautiful number, the biggest you’ve ever seen, and your limited partners need to pay you a fee just to manage it. Venture capital hasn’t seen valuations like this in a long time, and on paper, it feels like a lot of VCs got in on companies worth billions of dollars. On paper, Cognition is worth $10.2 billion, Perplexity $18 billion, Cursor $29.3 billion, Lovable $6.6 billion, Cohere $6.8 billion, Replit $3 billion, and Glean $7.2 billion — massive valuations for companies that all basically do products that OpenAI or Anthropic or Amazon or Google or any number of Chinese companies are already working to clone. They are all losing tons of money and have no path to profitability. Venture capitalists would, of course, claim I’m insane, saying that the “growth is obviously there” while pointing to whatever startup has made $100 million ARR ($8.3 million in a month), all while not discussing the underlying operating expenses. The idea, I believe, is that the current spate of AI spending is only set to increase next year, and that will…somehow lead to fixing margins? Venture capitalists staunchly refuse to learn anything other than “invest in growth and then profit from growth,” even if “profiting from growth” doesn’t seem to be happening anymore. In reality, venture capital shouldn’t have touched LLMs with a fifteen foot pole, because the margins were obviously, blatantly bad from the very beginning. We knew OpenAI would lose $5 billion in the middle of 2024. A sane venture capital climate would have fucking panicked, but instead chose to double, triple and quadruple down. I believe that massive valuation drawdowns are a certainty. There are losses coming. Venture capitalists, I have to ask you: what happens if OpenAI dies? Do you think that this will make investors interested in funding or acquiring other AI startups? How much longer are we going to do this? When will venture capital realize it’s setting itself up for disaster? And what, exactly, is the plan? OpenAI and Anthropic will suck the lakes dry like an NVIDIA GPU named after Nancy Reagan. How is this meant to continue, and what will be left when it does? The answer is simple: there won’t be money for venture capital for a while. Those AI holdings are going to be worth, at best, 50%, if they retain any value at all. Once one of these startups die, a panic will ensue, sending venture capitalists scrambling to get their holdings acquired, until there’s little or no investor interest left.
December 29, 2025 at 5:32 PM

by Brooke Harrington

24%
ebharrington.bsky.social
Old enough to remember multiple iterations of the AI hype cycle going back to the 1980s: it always ends like this, with the not-so-bright CEOs being fleeced by the tech hucksters, who slink away with their pockets full & destruction in their wake.

They don't teach this in b-school, but they should.
carnage4life.bsky.social Dare Obasanjo @carnage4life.bsky.social · 1d
I think companies are overestimating how much AI can improve productivity of entire teams but underestimating at the individual level.

Separate studies from both Forrester and Orgvue found 55% of employers who laid off workers because of AI now regret that decision. A “talent doom loop” indeed.
In defense of junior staff: Why replacing young people with AI could spark a 'talent doom cycle'
"You're missing a significant aspect of growth if you shut down the pipeline on bringing junior or entry-level talent into an organization," one CEO said.
www.cnbc.com
December 29, 2025 at 4:51 AM

by Guido Baldi

18%
gbaldi.bsky.social
AI boom or bubble? That's the question!

Robin Harding's opinion piece is convincing (no bubble, just a boom). But we should remain vigilant. A boom can easily develop into a bubble. The boundaries between boom and bubble are difficult to draw in practice.

www.ft.com/content/f229...
The AI boom is not a bubble
Valuations may be spectacular and a bust could come — but while there’s exuberance there is no mania or irrationality
www.ft.com
December 29, 2025 at 2:28 PM

by John Appleby

8%
jappleby123.bsky.social
Absolutely with @policyskeptic.bsky.social on the WhatTheActualFuck of this news piece on AI tool predicting A&E demand! I presume it has/will be evaluated? #notholdingbreath
December 29, 2025 at 6:02 PM

Reposted by Greg Linden

5%
edzitron.com
More than half of VC funding goes to AI companies that pay OpenAI and Anthropic, who then pay the hyperscalers, and at some point somebody is going to run out of money to pay somebody else. When that happens, the bubble bursts.

www.wheresyoured.at/the-enshitti...
OpenAI is trying to raise as much as $100 billion in funding so it can continue to pass money to one of a few public companies — $38 billion to Amazon Web Services over seven years, $22.4 billion to CoreWeave over five years, and $250 billion over an indeterminate period on Microsoft Azure. If successful, OpenAI’s venture telethon will raise more money than has ever been raised in a single round, draining funds that actual startups need. Anthropic has agreed to $70 billion in compute and chip deals across Google, Amazon and Broadcom, and that’s not including the Hut8 compute deal that Google is backing. This money will come from what remains of venture capital, private equity and hyperscaler generosity. Yet elsewhere, even the money that goes to regular startups is ultimately being sent to hyperscalers. That AI startup that needs to keep raising $100 million in a single round isn’t sending that cash to other startups — it’s mostly going to OpenAI (Microsoft, Amazon, CoreWeave, Google), Anthropic (Google, Microsoft, Amazon), or one of the large hyperscalers for Azure, AWS or Google Cloud. Silicon Valley didn’t birth the next big tech firm. It incubated yet another hyperscaler-level parasite, except instead of just spending money on hyperscaler services (and raising money to do so), both Anthropic and OpenAI actively drain the venture capital system as well, as they both burn billions of dollars. By creating something that’s incredibly expensive to run, they naturally create startups more-dependent on the venture capital system, and the venture capital system has no idea what to do other than say “just grow, baby!” Both OpenAI and Anthropic’s models might be getting cheaper on a per-million-token basis, but use more tokens, increasing the cost of inference, which in turn increases the costs of startups doing business, which in turn means OpenAI, Anthropic, and all connected startups lose more money, which increases the burn on venture capital. This is a doom-s…
December 29, 2025 at 5:32 PM

Reposted by Daniel Katz

10%

Reposted by Aric Rindfleisch

4%

by John Quiggin

37%
johnquiggin.bsky.social
What is the role of AI here? Bogus photo-shopped history has been on FB every since it existed.
www.reuters.com/article/fact...
Fact Check: Image purporting to show Time magazine cover warning of an impending ice age is digitally altered
An image purporting to show a Time magazine cover warning of an impending ice age is digitally altered.
www.reuters.com
December 29, 2025 at 2:07 AM

by Paul Nightingale

18%
paulnightingale.bsky.social
Just a quick reminder: AI hype (guff about singularities etc) is a coordinated attempt to change the subject and avoid talking about regulation.
December 29, 2025 at 9:18 AM

Reposted by Aleksandra Urman

6%
natematias.bsky.social
One lesson from the shocking story of this teenager's final AI interactions is how sensitive the data is. Journalists were not allowed to review the content itself.

Society shouldn't have to wait for deaths & lawsuits to pile up. We need faster science. (1/n)

www.washingtonpost.com/technology/2...
74 suicide warnings and 243 mentions of hanging: What ChatGPT said to a suicidal teen
Analysis of high-schooler Adam Raine’s ChatGPT account by attorneys for his parents shows how the chatbot became a confidant as he planned to end his life.
www.washingtonpost.com
December 29, 2025 at 2:26 PM

Reposted by Brigitte Nerlich

3%
fullfact.org
Full Fact’s Head of AI, @mrdudders.bsky.social, reflects on how AI has shaped misinformation this year—and how our own AI tools are helping fight back.
December 29, 2025 at 3:07 PM
Lightnews Lightnews Lightnews