Morning Edition

The Futurist

Friday, November 06, 2026
AI & Technology Markets & Crypto Ideas Worth Keeping
S&P 500$7,688.62▲ 1.59%
Dow Jones$53,356.12▲ 3.05%
Nasdaq$25,736.51▲ 1.46%
VIX$16.87▲ 1.81%
BITCOIN$80,435.00▲ 2.47%
RIPPLE$1.32▲ 2.53%
ETHEREUM$2,417.85▲ 3.87%
DOGECOIN$0.08▲ 2.34%
01

AI & Technology

404 MEDIA

Everyone Everywhere Hated Meta's New AI Training Terms. Now, They're On Pause

Meta's proposed commercial AI training terms triggered a rare unified backlash from developers, creators, and privacy advocates alike. The company has paused the rollout and is heading back to the drawing board — a significant retreat for a platform betting everything on AI dominance.

A16Z

Why Frontier Models Still Can't Parse a PDF

Reducto's CEO argues that AI agents are bottlenecked not by reasoning ability but by document parsing — PDFs, scans, and spreadsheets still defeat even the best models. Until the document layer is solved, consequential work in healthcare, legal, and finance stays out of reach.

A16Z

Cybersecurity in the Agentic Era

AI can now turn a patched vulnerability into a working exploit faster than most security teams can read the advisory. Traditional defenses built for human attackers and static malware are structurally unprepared for autonomous agents probing at machine speed.

ALL-IN

Anthropic S-1 Shocker, Greatest Business Ever? Google's $500B Wipeout

The besties tear apart Anthropic's S-1 filing, debating whether it represents the greatest business ever built or a speculative house of cards. Google's single-day $500 billion market cap wipeout provides a sobering backdrop to the AI euphoria.

GREG ISENBERG

SaaS is Dead?

Isenberg argues the per-seat SaaS model is structurally obsolete now that frontier models make tool-building trivial. The replacement: give the tool away free, harvest data and network effects, then monetize through AI agents and human services.

SIMON WILLISON

datasette-agent 0.5a0

Willison ships a significant Datasette Agent release featuring a persistent "panel" that lets an AI chat window sit alongside live data tables and queries. It's a quiet but concrete example of agents being woven into real working environments rather than demo sandboxes.

02

X / Twitter Signal

Mollick pushed back hard on AI doomer narratives tied to stock volatility, writing that any financial downturn in AI equities will not meaningfully alter the technology's trajectory. It's a useful corrective in a week where Google shed $500 billion in market cap — market pain and technological momentum are different phenomena running on different clocks.

Texas Governor-elect Gina Hinojosa called for a halt on AI data center development pending new legislation — a significant political signal in the state that has become ground zero for hyperscale compute buildout. Combined with diesel inventories hitting their lowest level since 2003, the energy and infrastructure constraints around AI are no longer theoretical.

OpenAI launched Images 2.6 with GIF support and faster generation, then confirmed an acquisition of code-hosting startup Pierre — two moves in one day that signal OpenAI is aggressively expanding its surface area beyond chat into creative tools and developer infrastructure simultaneously.

Google rolled out Gemini Omni Flash 1.2 with longer video generation, deeper cinematic camera controls, and broader multimodal support. Demis Hassabis amplified the announcement — the speed of iteration in the model-plus-media stack is now measured in days, not quarters.

03

The Thread

Two stories this week expose a deepening tension at the heart of the AI industry: the race to capture data and the growing resistance to that capture. Meta's forced pause on its AI training terms isn't just a PR stumble — it's evidence that the implicit social contract around data use is being renegotiated in real time, with users and regulators finally holding some cards. The backlash was fast, global, and effective. That's new.

The bottleneck isn't intelligence. It's the messy, analog world that intelligence has to operate in.
The more important structural story is infrastructure fragility. AI data centers are straining diesel supply chains, triggering political opposition from Texas to Brussels, and creating single points of failure that the freight and energy markets are only beginning to price in. When a governor-elect's first major policy stance is a moratorium on compute construction, the "endless growth" narrative for hyperscalers deserves serious scrutiny — independent of what any AI model can actually do.

Meanwhile, the software layer is being quietly rewritten. SaaS as a pricing model is cracking, document parsing remains a stubborn unsolved problem blocking real enterprise deployment, and cybersecurity assumptions built for human-speed attacks are already obsolete. The a16z cybersecurity conversation and the Reducto PDF argument point to the same underlying truth: the gap between what AI demos can do and what AI can reliably do in production is still vast, and closing it is where the next wave of real company-building happens.