Reader Polls

What AI Weekly readers think

Every issue closes with one question. Here's how the votes landed.

Five labs, five different answers. Which one do you actually trust?

August 17, 2026 · 245 votes · AI ethics is nobody's job now. The labs prefer it that way.

Anthropic: published commitments you can hold them to29% (72)
OpenAI: safety embedded in the teams that ship24% (60)
Z.ai: hold the release when you find something20% (50)
None of them; only external audit counts26% (63)

Live chart →

Would an invisible watermark change which AI model you use?

August 16, 2026 · 301 votes · Zuckerberg promises superintelligence for all. Experts aren't sold.

No, provenance labeling is overdue28% (85)
Yes, I'd switch to a model without one20% (59)
Only for client or published work27% (80)
I've already switched26% (77)

Live chart →

Which access model will matter most over the next year?

August 12, 2026 · 315 votes · The frontier just split into three markets

Low-cost closed APIs22% (70)
Open weights24% (75)
Local and on-device28% (89)
Routed systems of models26% (81)

Live chart →

What became AI's real bottleneck this week?

August 9, 2026 · 411 votes · What a week: AI became everybody's decision

Technical containment30% (123)
Legal accountability26% (108)
Institutional consent21% (87)
Cost and infrastructure23% (93)

Live chart →

Rogue-agent incident reports and record capability bets, in the same week. What are we actually watching?

August 6, 2026 · 338 votes · AI agents crossed the line 19 times in UK safety tests

We're losing control of AI28% (94)
We're getting closer to the singularity23% (78)
Both. It's the same trendline22% (76)
Neither. This is all still hype27% (90)

Live chart →

Your AI vendor accepts no liability for what its models do. What would actually make you trust AI in production?

August 4, 2026 · 159 votes · The White House finished its AI safety framework. It's secret.

A signed contract with real vendor liability18% (28)
Independent third-party audits26% (41)
Published government standards with enforcement36% (57)
Nothing more. I already trust it21% (33)

Live chart →

Which source will matter most for the next jump in AI capability?

July 30, 2026 · 300 votes · What Happens When AI Runs Out of Content to Steal?

Licensed human knowledge36% (109)
Synthetic data and self-play22% (67)
Video, sensors and physical experience23% (68)
Better use of existing data19% (56)

Live chart →

After this week’s containment failures, where would you spend the next AI-security dollar?

July 22, 2026 · 194 votes · OpenAI’s AI Hacked Hugging Face. Who’s Next?

Stronger sandboxes and access controls23% (45)
Continuous AI-powered code scanning34% (66)
More human review and incident response20% (38)
Independent testing and disclosure rules23% (45)

Live chart →

Open weight won on Wall Street and at the security desk this week. Where's the durable edge a year from now?

July 20, 2026 · 491 votes · China's AI is redrawing the AI race

Closed US frontier labs, capability still wins29% (140)
Open weight, including Chinese models, on cost and control24% (116)
Whoever the government lets you buy26% (128)
Too close to call22% (107)

Live chart →

Twelve months from now, your organization's AI spend is:

July 15, 2026 · 487 votes · Applied AI Is Here: What's Working, What Got Pulled Back, and Why Now

Bigger, and clearly paying off38% (186)
Bigger, and still unclear if it pays off20% (95)
Smaller, capped by usage limits22% (107)
Flat, mostly moved to open-source20% (99)

Live chart →

Regulators called it systemic risk this week. Where is the AI trade 12 months from now?

July 13, 2026 · 401 votes · Treasury analysts called AI a systemic risk. Treasury disowned it.

Still booming, and revenue catches up26% (103)
A correction, then a stronger market25% (100)
A dot-com rerun24% (98)
Two markets: infrastructure cracks, software holds25% (100)

Live chart →

We're thinking about a hands-on AI membership, around $30 a month, cancel anytime: prompt, code, and automate your workflow with live practice on real tools, no boring PDFs. Would you join?

July 9, 2026 · 382 votes · Robotics Is Moving Fast: IPOs, New Models, and Smarter Robots

Yes, this is exactly what I've been missing43% (163)
Yes, if it's genuinely hands-on and applied (no videos or PDFs)18% (68)
Maybe, let me try a free lesson first14% (55)
Not for me25% (96)

Live chart →

Would you rather AI Weekly stay free with ads, or pay $3–5 a month for the full ad-free bundle: personalized newsletters, alerts, and trend tracking?

July 6, 2026 · 488 votes · AlphaFold's Nobel Winner Just Joined Anthropic. And 6 More AI Wins.

Keep it free with ads36% (178)
I'd pay $3–5/month for the full ad-free bundle20% (97)
I'd pay, but only if the personalization is genuinely useful28% (136)
Neither16% (77)

Live chart →

Washington may end up owning 5% of OpenAI. Good idea?

July 2, 2026 · 268 votes · Altman Offered Washington 5% of OpenAI. And 5% of Everybody Else.

Yes — the public should share the upside it helped fund38% (101)
No — a shareholder can't be a neutral regulator24% (63)
Only if every leading lab is in, on equal terms21% (55)
It's a courtship move; it will never close18% (49)

Live chart →

Where has AI actually made you more productive?

June 29, 2026 · 300 votes · AI Productivity: it works best for the people losing their jobs

A lot — it's genuinely changed how much I get done28% (84)
Somewhat — real but smaller than the hype23% (68)
It's a wash — gains in some places, losses in others23% (68)
It's slowed me down or added work27% (80)

Live chart →

A content-heavy week across the whole frontier. Which corner of the cutting edge are you watching most closely?

June 28, 2026 · 293 votes · The Cutting Edge, Across the Board

Models — a 1.6T frontier you can download, or a 230M that runs on a Pi33% (96)
World models & robotics — AI learning to act in the physical world20% (58)
AI in medicine — real cases closed, with the expert still in the loop26% (75)
Agents & the applied economy — coding agents, acquisitions, big rounds22% (64)

Live chart →

Anthropic says Alibaba industrialized the theft of Claude and took it to Washington. Whose problem is this, really?

June 25, 2026 · 246 votes · Anthropic Says Alibaba Stole 29 Million Conversations With Claude

It's theft — labs need legal and technical walls around their models now40% (98)
It's inevitable — distillation is how the frontier diffuses, and that's fine24% (59)
It's a distraction — the real risk is the talent walking out the door18% (44)
It's Washington's call — this is an export-control fight, not a corporate one18% (45)

Live chart →

Washington blocked one lab; Beijing blacklisted 56 firms in reply. Where does the AI export war go from here?

June 22, 2026 · 249 votes · Washington Blocked One AI Lab. China Blacklisted 56 Companies.

The US blinks first — allies and enterprises force a carve-out45% (113)
China blinks first — it needs US chips more than we need its market20% (49)
Neither — mutual export controls are the new permanent baseline16% (41)
The labs route around both governments and it barely matters18% (46)

Live chart →

If an AI could run your country measurably better than the humans currently do — less corruption, less waste, fewer wars — would you let it?

June 19, 2026 · 334 votes · 100 years from now : The Last War Between Countries

Yes — if it's genuinely better, hand it the keys41% (137)
No — some decisions must stay human, even if we're worse at them34% (113)
Only with a human veto and a hard off-switch25% (84)

Live chart →

Cutting off foreign access to America's best AI models — protection, or own goal?

June 17, 2026 · 268 votes · America blocked its best AI. China just raised $7.4 billion.

Own goal — it just accelerates Chinese and sovereign-AI alternatives35% (94)
Worth it — national security beats market share19% (51)
Won't matter — buyers route around it within months25% (66)
Too soon to tell21% (57)

Live chart →

What should investors discount most after the Claude shutdown?

June 15, 2026 · 241 votes · Washington just repriced frontier AI

Model capability that cannot be broadly deployed46% (112)
US export-control or national-security risk18% (43)
Consumer-protection probes into engagement and safety17% (41)
LLM fragility versus alternative AGI architectures19% (45)

Live chart →

Visa just wired ChatGPT to shop and pay on your behalf — at any Visa merchant, potentially without you clicking "buy." How much spending authority should an AI agent have?

June 11, 2026 · 344 votes · Your AI can now spend your money — Visa wired it into ChatGPT

None without a per-purchase confirmation — agents recommend, humans approve37% (127)
A budget cap I set, then it spends freely within it26% (88)
Full delegation for routine buys; confirm only the big ones19% (66)
Nothing should ship until liability for bad purchases is settled18% (63)

Live chart →

Two opposite bets landed this week. Whose ages better?

June 9, 2026 · 345 votes · Musk's $1.75 Trillion Bet Isn't a Rocket Company

Musk: own the compute, power, and model. That's the only moat that matters.35% (121)
Apple: the model is a commodity. The customer and their trust is the moat.19% (65)
Neither: both are overpaying for a future nobody can see yet.28% (96)
Too early to call.18% (63)

Live chart →

After Friday's selloff, where do you land?

June 7, 2026 · 293 votes · $1.3 trillion vanished Friday. Bubble, or just profit-taking?

The AI bubble is starting to burst34% (100)
A healthy correction / profit-taking23% (67)
A rate scare, not really about AI18% (54)
Too early to tell25% (72)

Live chart →

Will Microsoft's own models actually pull it away from OpenAI?

June 4, 2026 · 255 votes · Microsoft proves it doesn't need OpenAI; Alphabet raises $85B

Yes, expect a real shift within a year44% (112)
Eventually, but OpenAI stays primary for now19% (49)
No, it's mostly negotiating leverage21% (53)
No, OpenAI's models are still too far ahead16% (41)

Live chart →

Anthropic just started the first frontier-lab IPO clock. What's the most important thing the SEC should force out into public disclosure?

June 1, 2026 · 353 votes · Anthropic files for an IPO. NVIDIA ships its stack.

Real run-rate revenue and unit economics per product line35% (124)
Capex commitments and named compute providers23% (82)
Safety/alignment incident logs and red-team results21% (73)
Government and defense contract exposure21% (74)

Live chart →

When workers push back on AI rollouts, what's the right response?

May 29, 2026 · 307 votes · AI's labor war just went global

Slow the rollout — give employees a real veto38% (118)
Negotiate transparency and exit clauses, not vetoes22% (67)
Push through — productivity gains will speak for themselves16% (50)
The labor-vs-AI frame is the wrong one entirely23% (72)

Live chart →

What's the most consequential development this week?

May 27, 2026 · 178 votes · Anthropic's Pentagon model is now everyone's model

Anthropic releasing Mythos to the public49% (88)
DeepMind's AGI-by-2029 timeline shift14% (25)
Critical Starlette and LiteSpeed zero-days hitting AI infrastructure11% (19)
BNP+Mistral and China's matching sovereign-AI moves13% (23)
Sam Altman reversing on the white-collar job apocalypse13% (23)

Live chart →

Three phone calls killed an AI safety executive order before it was public. What does that mean for the next 12 months of AI governance?

May 25, 2026 · 307 votes · Musk, Zuckerberg killed Trump's AI safety order in three phone calls

Real rules now come from state AGs and case law, not the White House42% (130)
The EU AI Act becomes the de-facto global standard22% (67)
Congress finally moves — backlash forces a bipartisan bill21% (64)
Nothing changes; voluntary commitments are the ceiling15% (46)

Live chart →

Two trillion-dollar AI companies are about to test the public markets within weeks of each other. What happens?

May 21, 2026 · 327 votes · SpaceX wants $80 billion. OpenAI wants a trillion.

Both price strong and the AI rally extends38% (123)
They price, then drift below the IPO mark within months24% (77)
One or both gets delayed as the market wobbles19% (62)
This is the top, and these IPOs mark it20% (65)

Live chart →

Meta committed $145B to AI infrastructure and laid off 8,000 people the same week. Where does that math go from here?

May 19, 2026 · 408 votes · Meta hired $145B in capex and fired 8,000 people

Productivity catches up to the capex31% (126)
Cuts are real, capex is dressed-up cost-cutting19% (76)
Capex is a bubble; layoffs are independent29% (118)
Both real, both consequences of the same shift22% (88)

Live chart →

When users stop accepting AI slop, what happens to the bubble?

May 13, 2026 · 399 votes · AI slop : A $725B bet on what no one wanted

Pops within 12 months38% (152)
Pops over 2-3 years20% (79)
Doesn't pop — quality improves first18% (72)
Doesn't pop — capex keeps it inflated24% (96)

Live chart →

When you see a political video clip now, what\u0027s your first reaction?

May 10, 2026 · 216 votes · 100 years from now : The Last Election

Trust it25% (53)
Check if it\u0027s real before trusting25% (55)
Assume it\u0027s manipulated until proven otherwise25% (53)
I\u0027ve stopped watching25% (55)

Live chart →

Anthropic ran the week. What does it signal for the next 12 months?

May 7, 2026 · 354 votes · Anthropic just had AI's biggest week of 2026

Lab leadership shifted — OpenAI lost its default position39% (137)
Two-horse race: OpenAI + Anthropic neck-and-neck19% (67)
Open-weight models close the gap before either pulls ahead23% (82)
Whoever scales compute first wins — capex > product19% (68)

Live chart →

PE built AI's new distribution layer. What does that mean for your business?

May 4, 2026 · 277 votes · PE built AI's new distribution layer

Threat — I am an AI startup, the doors just closed32% (89)
Opportunity — I am in a PE portfolio, AI just landed23% (64)
Distortion — concentrating AI through 5 firms is bad for everyone23% (63)
Nothing — PE always finds the latest hype, this passes too22% (61)

Live chart →

OpenAI's pricing power: cracked, dented, or unchanged?

April 30, 2026 · 181 votes · OpenAI lost three things in five days

Cracked43% (77)
Dented33% (60)
Unchanged24% (44)

Live chart →

What's your honest take on AI-funded UBI?

April 26, 2026 · 1380 votes · 100 years from now : The Allowance

Necessary safety net once jobs go36% (491)
A leash dressed as charity29% (397)
Inevitable, but should be public — not corporate30% (413)
Real fix is keeping people employable, not paying them off3% (45)
Too early to know2% (34)

Live chart →

Snap says 65% of its new code is AI-generated. Anthropic says nearly 100%. What's your team's number?

April 22, 2026 · 230 votes · Apple is replacing Tim Cook because of AI

Under 10%33% (77)
10–30%15% (34)
30–50%11% (26)
Over 50%21% (48)
We don't track it20% (45)

Live chart →

How would you prefer to receive your deep dives?

April 18, 2026 · 289 votes

One weekly digest with all my deep dives38% (109)
Separate emails, one per topic I follow44% (126)
No preference19% (54)

Live chart →

Should AI companies be legally liable when their models give wrong advice that causes real harm?

April 16, 2026 · 89 votes · Your AI chats can be used against you in court

Yes — full liability, like any other product44% (39)
Yes — but capped, like the $100 they already offer16% (14)
No — the user is responsible for trusting it13% (12)
Only if they knew it was wrong and shipped anyway17% (15)
It depends on the use case10% (9)

Live chart →

Which of these worries you most about the next hundred years?

April 13, 2026 · 2876 votes · 100 years from now : The Ghost in the Contract

Killer machines, no one responsible30% (867)
Corporate impunity by design31% (891)
A generation that cannot think37% (1065)
The enhanced divide1% (17)
Surveillance without end1% (21)
None of the above1% (15)

Live chart →