🚨 AI-Related Security Incidents Are Surging
AI-related security incidents jumped from just 24 in Q1 2026 to a record 169 in Q3 - a 7x increase in two quarters.
With 89 incidents already recorded in Q4*, the rise in AI-driven cyber threats is becoming impossible to ignore.
As AI adoption accelerates, is cybersecurity keeping up?
1
17
How to build an AI agent token: the dos and don'ts.
Rule zero: no agent, no token.
AI agent tokens that lost 90%+ in 2025 almost all launched narrative-first, with nothing functional behind them.
Thread.
1
1
1,963
The Don'ts
- Launching a token for an agent that doesn't exist yet
- Unlocking on a calendar. AI16Z to ELIZAOS took supply from 6.6B to 11B. Token fell 28% in 24 hours
- Auditing only the contracts. Only 29% of agentic AI teams report security readiness
The Dos
- Run the agent 4 to 8 weeks on testnet and measure real output first
- Tie every emission to agent activity. Akash burns $0.85 of AKT per $1 of compute
- Vest linearly. Team allocation 15 to 20% on a 4-year vest
- Announce supply changes 30 days ahead
1
1,881
The part most teams skip: the agent layer.
A contract audit misses credential hoarding, prompt injection and one breached orchestrator exposing every downstream agent. Barracuda found 43+ agent framework components with embedded exploits.
Review four surfaces: contracts, credentials, agent-to-agent comms, framework supply chain. Budget $10K to $60K for the audit and $500 to $5K a month for compute.
The token is a receipt for agent work. Whether that receipt is worth anything depends on how well the agent is secured.
104
🚀 Gemini 4 Neon is raising the bar for AI performance.
Google's latest model leads across multiple benchmarks, outperforming Claude Opus 5.5 and GPT-6.2 Sol in reasoning, agentic coding, knowledge work, and long-context tasks.
📊 95.2% on GraphWalks. 90.5% on SEC-bench Pro. 84% on RiemannBench.
The AI race is no longer just about smarter models. It's about agents that can execute complex tasks at scale.
The next era of AI is agentic.
🚨 Anthropic makes history with a record-breaking $81.3B IPO!
The company behind Claude reportedly debuted on Nasdaq under $ANT at a staggering $1.5T valuation.
📈 IPO price: $83
🚀 Closing price: $97.17 (+17%)
💰 400+ employees reportedly worth over $100M each
AI isn't just reshaping technology. It's rewriting the rules of Wall Street.
But the trillion-dollar question remains: Can AI giants turn explosive growth into sustainable profits?
AI of the week: Talus @Talus_Labs
Trustworthy AI agents on Sui.
Agents get on-chain identity, scoped permissions, and verifiable workflows - so they can handle real value without you losing control.
Protocol v2.0 is live on mainnet.
On-chain proofs + failure handling + priority execution.
This is the infrastructure layer for agents that actually get trusted with capital.
Most agent platforms focus on launching tokens or chat interfaces.
Talus is building the execution + trust stack:
• Talus Agent Package (identity + capabilities)
• On-chain workflow DAGs with scoped grants
• Verifiable execution proofs
• Per-step settlement and risk controls
• Accelerator backed by Sui Foundation
Agents only scale when people are willing to delegate real money.
Talus is one of the few building the rails that make that possible.
2,303
Trade Base assets directly from ChatGPT.
@ChatGPT now connects to your Coinbase account, so you can buy, sell, and manage Base tokens right inside the chat.
The future of onchain trading is getting conversational.
teapaw.org/base/status/2122694622…
Trade Base assets from ChatGPT
@ChatGPT now connects to your Coinbase account, so you can trade Base assets in chat
🚀 GPT-6.2 Astra just landed with an Epoch Capabilities Index (ECI) of 174.
That puts it on par with Fable 5.2 and above the reasoning-era frontier trend of about +16 ECI/year. 📈
What stands out:
• The frontier has climbed steadily since o1, and Astra sits at the very top of the curve
• It lands right at the edge of the trend’s prediction band, not far beyond it
• The race between OpenAI and the Fable line is basically neck andneck
Is this the start of a new acceleration, or just the trend doing its thing? 👇
Source: Epoch AI
Anthropic’s IPO prices tonight. First trade on Nasdaq tomorrow ($ANT).
The setup:
• 825M Class A shares at $70 - 78
• Roadshow launched Nov 10, pricing expected Nov 18, first trade Nov 19
• About $916B - $1.02T at the range on shares outstanding (fully diluted estimates run $1.3T - $1.4T)
• Up to 20% earmarked for retail, though allocation depends on demand
Crypto platforms are also offering tokenized exposure ahead of the listing. Terms vary, fills aren’t guaranteed, and these products may track the price without making you a shareholder.
The final price can also land outside the range.
Elon: ‘Blue check coming back for free on the house" 🤣
CZ: money is rotating into the Anthropic IPO.
Short-term pressure on crypto. Longer-term, the AI x crypto overlap is still expanding.
teapaw.org/cz_binance/status/2122…
Why are crypto prices down?
No one knows for sure. Correlation does not prove causation. Here are some potential reasons.
Money shifting to the Anthropic IPO: teapaw.org/ColeMacro/status/21212…
ETF outflows: U.S. spot Bitcoin ETFs saw about $1.9 billion in outflows over seven trading sessions through November 13. Ether funds also saw withdrawals. crypto.news/bitcoin-price-fa…
Security fears: the Binance hack, iPhone-linked crypto wallets drained, Chrome exploit, Microsoft down. These security incidents could make some people less willing to keep funds in crypto.
Hot U.S. inflation has raised expectations of another Fed rate hike. Higher yields and a stronger dollar can weigh on "riskier assets" such as crypto.
Political uncertainty: Midterm results have raised concerns about potentially stricter U.S. crypto policy.
No, Binance did not sell crypto.
I bought crypto in the past few days, as I got some stablecoins recently. (maybe this is the cause 🤣)
Mid-long term, crypto will continue to grow.
Views are my own. Not financial advice.
For future reference, price of BTC hit a low of $62,006 today.
🔥 Latest AI x Crypto / DeFAI Developments:
• @lightning Labs just shipped pi402. Coding agents can now pay for AI inference per request in Bitcoin over Lightning (via Wavelength). No accounts, no API keys, no cards. Demo pricing ~$0.375 / 1M input tokens. Agents are starting to buy their own compute.
• x402 keeps scaling. Lifetime volume has crossed ~200M transactions and ~$43M settled across 12 chains. October was quiet (~5.1M txs / $274K), but November is running hotter. Still mostly sub-$1 micropayments. @base remains the clear leader. @x402 foundation is live under @linuxfoundation.
• Identity layer continues to grow. ERC-8004 registries now show ~2.2M agents registered across 24 chains (@base still #1 with ~1.19M). Most are still placeholders — real active agents remain a much smaller number.
• Narrative vs reality: AI Agents narrative sits at ~$2.5B market cap, DeFAI at ~$450–490M. Micropayments and tool use are real. Autonomous capital deployment at scale is still early.
Agents are getting better wallets, better payment rails, and clearer identity standards. The bottleneck has shifted from “can they act?” to “can they be trusted with real capital?”
Payment rails are here. Reliable judgment is still the missing piece.
Before AI: a full pyramid of humans.
After AI: the bottom layers quietly become agents.
teapaw.org/python_academia/status…
AI agents can now search the web and pay onchain. 🤖
Brave Search API + Base + x402 = pay-as-you-go access with USDC.
No subscriptions. Just autonomous agents making payments.
The agentic economy is getting real. 🔥
teapaw.org/base/status/2121349892…
Learn more: teapaw.org/base/status/2121349893…
3
2,206




















