Oct 2, 2026 · 1h 6m
Crypto infrastructure faces tests from hacks, incentives, and AI
The Chopping Block: Bitget's 387 Million Dollar Hack, Kalshi's Cooked Perps Volume, and Agentic Bank Runs
The episode connects three emerging sources of financial risk: discretionary cross-chain intervention, opaque market incentives, and capital that can move automatically.
- 1Cross-chain protocols must balance censorship resistance against the need to freeze or block assets linked to theft.
- 2Validator failures and Ethereum’s implicit governance complicate the idea that staking offers a risk-free return.
- 3AI agents could turn sluggish bank deposits into mobile capital, accelerating both yield migration and coordinated withdrawals.
Don't miss
The discussion imagines AI agents automatically moving deposits toward higher yields, potentially making bank runs faster and more coordinated.
The brief
A roughly $387 million Bitget hack, reportedly linked to North Korea’s Lazarus Group, becomes a test of whether cross-chain infrastructure should stay neutral or block suspicious flows.
The comparison between THORChain and Near Intents sharpens the central tension: censor-resistant systems preserve neutrality, while intervention can help enforce property rights after theft.
A Consensys validator compromise shifts the debate from lost rewards to Ethereum’s governance, asking whether staking risk includes downtime, slashing, and possible rollback decisions.
Kalshi’s unusually high volume and alleged incentive-driven repetitive trading raise a basic market question: when does subsidized activity stop being useful liquidity and become misleading signal?
The episode’s most consequential scenario is agentic banking: software that automatically chases yield could drain deposits faster than banks can respond, creating automated bank runs.
What was said on this episode
28 statements · 12 positive · 11 negative · 5 neutral
Bitget ranks among the world’s top five exchanges by derivatives volume.
“BitGet is one of the biggest exchanges in the world, top five in any given day with respect to derivatives volumes.”
Listen at 1:48
The Bitget hack was likely attributable to North Korea’s Lazarus Group.
“very likely attributable to the Lazarus Group. which is a part of North Korea”
Listen at 2:00
Bitget’s $460 million protection fund was sufficient to cover all hack-related claims.
“they had $460 million set aside in an insurance fund slash user protection fund, so that was enough to fully pay out all the claims.”
Listen at 2:21
NEAR Intents is a cross-chain trading protocol.
“NIR also has this thing called NIR Intense, which is a cross-chain trading protocol.”
Listen at 4:30
NEAR Shield detects anomalies and reported hacks at the validator layer.
“NIR has a protocol called Shield. which sits at the validator layer, which basically looks for anomalies as well as looking for reports of hacks”
Listen at 4:39
NEAR Shield can refuse or stop transactions involving demonstrated high-risk parties.
“will refuse to trade with people or sometimes even stop people who are demonstrated to be high risk”
Listen at 4:50
The optimal user state combines low friction with very high protection.
“it's low friction but very high protection.”
Listen at 21:02
Long-tail chains have lower liquidity and make funds harder and slower to move.
“if you have money on. some like janky ass, you know, kind of long tail chain. There's less liquidity. It's hard to move your money out. Bridging takes longer.”
Listen at 21:47
Distressed assets may sell below their accounting par value.
“if you try to sell it, you're not going to get par”
Listen at 23:26
Distressed Railgun-held USDC would sell below par value.
“if you were distressed and you had to sell a bunch of USDC sitting in Railgun, you would not get part.”
Listen at 23:58
Cross-chain versions of the same asset should have small price spreads.
“No, it should be a small spread.”
Listen at 26:11
Researchers should study crypto asset premiums across security and convenience differences.
“people should actually study these premiums in crypto”
Listen at 26:54
AI agents will move users into better financial products.
“Will switch you to the better product.”
Listen at 54:24
People should not keep money in checking accounts earning only one basis point.
“Nobody should have money in a checking account anymore. The one basic point you're earning is insufficient.”
Listen at 54:50
AI agents will optimize users’ lives and portfolios.
“AI agents will optimize your life and your portfolio.”
Listen at 54:56
Higher deposit rates could retain funds during a JPMorgan bank run.
“if there is a bank run being experienced by JP Morgan, they just need to raise their rates and they'll keep the money.”
Listen at 55:14
Banking products rely on consumer inertia and laziness.
“The entire product is built on the laziness of consumers.”
Listen at 56:04
Banks keep deposit rates near zero because depositors are unlikely to leave.
“The reason why rates are zero is because. They have actual hard evidence that their depositors will not leave.”
Listen at 56:08
AI agents will overcome consumers’ inertia in managing deposits.
“I do think that agents will solve this for people.”
Listen at 57:56
Agent-driven deposit migration will occur gradually.
“this will be gradual.”
Listen at 58:02
Agent-driven banking changes will be gradual rather than sudden bank runs.
“It's not going to look like a bank run, which implies sudden, right? This is going to be a very gradual tectonic shift.”
Listen at 58:48
JPMorgan will strongly resist agent access to customer accounts.
“JP Morgan is going to fight that to the end of their days”
Listen at 59:13
Amazon’s business model would break if agents bypassed its advertising system.
“Amazon's business model breaks.”
Listen at 1:00:29
Muse recovered $200 in insurance overcharges for Tarun Chitra.
“I've got 200 bucks back so far.”
Listen at 1:01:56
Muse will cancel difficult-to-cancel consumer subscriptions.
“Muse will start canceling subscriptions that are hard to cancel”
Listen at 1:02:19
AI agents will eventually produce SVB-style bank runs.
“I do think we will also see actual agentic bank runs in the sense of like SVB style bank runs”
Listen at 1:02:48
Agent-managed capital could become a globally mobile $3 trillion pool.
“The whole world becomes the hot ball of money instead of like a $10 billion hot ball of money. It's a $3 trillion hot ball of money that just moves around.”
Listen at 1:03:17
AI systems will accelerate coordinated withdrawals during bank-risk events.
“It seems like AIs will do that even faster.”
Listen at 1:04:39
Statements are attributed to the speaker as said on the episode and reflect their view at the time, not PodLume's. They are not advice.