⚡ Why this matters

• MTurk trained the datasets that trained the models that replaced MTurk.
• The 46% number means the marketplace stopped being human before Amazon stopped selling it.
• This is the cleanest displacement story of the year - no interpretation needed.

🔍 What happened

• Announced Aug 25, 2026. Service ends Sept 30, 2026.
• New customer signups were cut off July 30, alongside SageMaker Ground Truth and Amazon Augmented AI (A2I).
• MTurk launched in 2005 to label data and do micro-tasks machines could not.
• A 2023 EPFL study estimated 33-46% of MTurk crowd workers used LLMs for text tasks.
• Amazon is shutting the whole human-in-the-loop labeling stack, not just the marketplace.

💬 Smart takes

• The labeling market did not die. It moved to expert vendors like Surge and Scale at 50x the price per task.
• Cheap generic human labor lost. Expensive specialist human labor got more valuable.
• Academic researchers lose their default subject pool. That is a quiet methodological problem.

🧭 Where this goes

1. Watch where the 500,000 workers go. Prolific and Clickworker are the obvious catchers.
2. Watch RLHF pricing. Expert annotation rates have been climbing all year.
3. Watch for replication crises in papers that used MTurk panels after 2023.

⚡ Why this matters

• Consent design is the whole story. The default decides the outcome, and everyone building these systems knows it.
• Amazon owns Twitch. The training data was already sitting inside the company, so nobody had to negotiate for it.

🔍 What happened

• Twitch added an account setting on August 12 that stops channel content from feeding Amazon's generative models.
• The toggle shipped switched on for every account. Streamers were enrolled before the option to decline existed.
• Twitch executives took questions on a Patch Notes livestream the same day.
• Almost 14,000 people upvoted a support-forum comment demanding the setting be opt-in.

💬 Smart takes

• CPO Mike Minton, asked why it is not opt-in: 'If this was opt-in, nobody would opt in. That's honestly the answer.'
• Head of community Mary Kish on why creators were not emailed: 'That's not the best answer, but creators don't always read their emails.'
• Minton confirmed back in 2024 that Amazon was already training on Twitch content, so the setting formalizes something old.

🧭 Where this goes

1. Likelyother creator platforms ship the same toggle in the same on position, now that someone has gone first.
2. Possiblea regulator reads 'nobody would opt in' as an admission and asks about the consent basis.
3. Wild Carda creators' collective builds a public opt-out registry and turns the default into a bargaining chip.

⚡ Why this matters

• First appellate precedent on whether AI agents may act on websites for their users. Every agent builder was waiting for this.
• The CFAA, the federal anti-hacking law, was site owners' main weapon against unwanted bots. This ruling blunts it for user-delegated agents.
• Agentic shopping and browsing just moved from legal gray zone to defensible.

🔍 What happened

• On Aug 4-5 the Ninth Circuit vacated Amazon's preliminary injunction against Perplexity's Comet browser and shopping agent.
• Holding: Amazon is unlikely to prove Perplexity accessed its servers under the CFAA.
• Reason: users operate Comet. It logs into the user's own Amazon account with stored credentials and buys through normal checkout.
• The panel applied the rule of lenity, reading the ambiguous statute against liability.
• The opinion notes there is little to no existing caselaw on ascribing responsibility for AI agents.

💬 Smart takes

• The Ninth Circuit panel: there is 'little to no existing caselaw directly dealing with how to ascribe responsibility for AI agents.'
• Search Engine Journal: framed it as the CFAA case that decides whether AI agents can visit your website.
• The Skeptic: a vacated preliminary injunction is not a merits win. Amazon still has contract and trademark angles, and this binds one circuit.

🧭 Where this goes

1. LikelyAmazon and other retailers rewrite terms of service to target agents through contract law instead.
2. LikelyOpenAI, Google, and Anthropic shopping agents expand faster now the hacking-law threat has faded.
3. PossibleCongress moves to update the 1986 hacking law with explicit AI-agent language.
4. Possibleanother circuit rules the other way, setting up a Supreme Court fight.
5. Wild CardAmazon flips and launches paid agent access for outside shopping bots rather than fighting them.

⚡ Why this matters

• AI demand is now visible in hyperscaler earnings, not just Nvidia's order book.
• Capacity, not model quality, is the constraint every AI roadmap inherits next.
• A $220 billion capex year resets what infrastructure spend means for the whole industry.

🔍 What happened

• Aug 3 - Amazon's market value topped $3 trillion for the first time.
• The shares posted their biggest one-day jump since April 2012.
• Second-quarter AWS revenue grew 37% to $42.2 billion, the fastest pace in 18 quarters.
• Total quarterly revenue crossed $200 billion for the first time.
• CEO Andy Jassy: the AI and chips businesses each run above $25 billion annualized.
• Capital spending guidance for 2026 now sits at $220 billion.

💬 Smart takes

• Andy Jassy, Amazon CEO: "AWS is booming... our AI and Chips businesses each eclipsed run rates of more than $25 billion."
• Tomasz Tunguz, Theory Ventures: reads the quarter as AWS finally answering the cloud-race question after years of trailing Azure's growth rate.
• Skeptic: $220 billion of capex only pays off if AI workloads keep growing into it - a demand wobble turns the buildout into overcapacity.

🧭 Where this goes

1. LikelyGoogle and Microsoft answer with higher capex guidance next quarter.
2. LikelyAWS keeps reaccelerating as enterprises consolidate AI workloads onto fewer clouds.
3. PossibleAmazon's Trainium chips take visible inference share from Nvidia by 2027.
4. Wild CardAWS reaches Jassy's $1 trillion revenue path far faster than the decade everyone assumes.