• Web3 on Fire
  • Posts
  • DeFi vs AI code, a death threat or an upgrade?

DeFi vs AI code, a death threat or an upgrade?

Also: Blackrock's biggest dump this year

GM, frens! ☕️

People have a long history of being suspicious of new technology. People once worried trains were too fast for ordinary life, electricity was dangerous, the internet was a toy for nerds. Every generation gets its turn to look at something new and confidently explain why it’ll never catch on.

Most of the time, the technology survives the criticism. The interesting part is figuring out which ideas are genuinely ahead of their time and which ones are just new for the sake of being new 🧠 

Here’s what we’re looking at this week:

  • 🤖 DeFi vs AI code, a death threat or an upgrade?

  • 🪨 Blackrock makes the biggest BTC DUMP of the year after poor performance

  • 🔥 Ethereum is “outgrowing” its Foundation and old narratives

  • 🤦‍♂️ Google sold the top search spot to a fake Uniswap wallet drainer (again)

Below is how $WALLET is trading right now.

Our Discord is always open if you want to argue about what looks ridiculous now and obvious later 🤠 

DeFi vs AI code, a death threat or an upgrade?

AI code has pretty much shook DeFi as of late and the issue has pretty much turned a long running security debate into a full blown industry argument.

The claim that followed brought every version of the argument out into the open pretty fast 👇️ 

  • Manuel Araóz, co founder of OpenZeppelin, went full alarm bell mode and said he now considers all of DeFi unsafe. You would think it’s about some risky fringe projects or “new protocols” or stuff like “a protocol with 900% APY and a frog logo” but no, he means all of it 🤔 

  • His argument is that AI agents changed the whole balance between defenders and attackers in cybersec. Security teams have to find every weakness. Attackers only need one. Now attackers have tools that can scan contracts faster, connect clues faster and keep testing the same target without getting tired or needing a coffee break 🥷 

That is a serious point. DeFi security was already a chaotic topic. Then when you also add AI…

  • OpenZeppelin itself recently said crypto companies lost more than $3.4 billion to hacks in 2025, though it also blamed most of that on compromised credentials, operational failures and code that changed between audits, rather than pure smart contract bugs 🤓 

  • Also as we discussed here previously, in April alone more than $650 million were stolen, including $292 million from Kelp and another $285 million from Drift after what experts described as “months of social engineering”.

So yes, the space has a problem. Nobody sane is arguing otherwise. But did AI coding and agents specifically made the whole thing doomed? Or whether they are just the next tool that forces everyone to grow up a little faster? 🤖 

Because every major technology does this.

Electricity was probably terrible news if your whole life was tied to steam engines. If you owned the machine, serviced the machine, funded the coal, trained people around the machine or built yourself some little kingdom around the machines, then yes, the future showed up looking EXTREMELY rude.

Still, nobody seriously looks back and says humanity should have kept the steam age, because we’re afraid to hurt boiler investors 🤡 

That’s why the AI coding argument is so heated. A new tool can absolutely break old assumptions and still be worth it in the long run

Aave Chan Initiative founder Marc Zeller pointed out that fewer than 10% of DeFi issues in the past year came from code level vulnerabilities, with most failures tied to infra, tracking back to bad risk setup, collateral management and weak operations 👇️ 

Polaris Finance dev Robert made a similar point, saying actual smart contract exploits are almost nonexistent these days, while recent breaches were mostly tied to centralized components that allowed human control rather than immutable onchain logic 🤨 

  • Vitalik has also written a good example not long ago, and it fits this case well. He said that AI assisted formal verification could make crypto systems more secure over time by helping devs write both code and mathematical proofs for that code 👨‍💻 

  • I.e AI does not only give attackers a sharper knife but can also give defenders better armor, assuming they use it properly and do not treat it like a magic spell 🧙 

The “vibe coding” meme is still annoying, though.

People talk about it like devs are just whispering “make DeFi safe” into a chatbot and pushing whatever comes out. If someone has no idea how code works, sure, that is how it’s going to work but it would be basically nothing more than gambling 🃏 

For actual devs, AI assisted coding is not replacing thinking. It is replacing a lot of the slow, repetitive grunt work. Any dev who has used these tools properly knows it is faster than coding alone. The problem is not the tool. The problem is pretending the tool removes the need for judgment 🤷‍♂️ 

That is probably where DeFi is headed, to be quite honest. AI agents will make crap teams more dangerous because they can produce slop faster without understanding more. But they can also make good teams stronger because serious devs can test more, review more, simulate more and verify more before anything touches user funds 🙃 

“AI ruined DeFi” is still just too lazy of a take. The truth, corroborated by plenty of people in the industry from what we can see, is that AI is exposing which parts of DeFi were already quite dysfunctional. Sloppy ops, centralization, low quality audits and teams treating security like an afterthought were problems before AI showed up. AI just makes the FAFO consequences arrive faster 🤐 

Blackrock makes the biggest BTC DUMP of the year after poor performance

The market wanted one clean week, because apparently asking for a normal stretch is now considered greed. Instead, the largest ETF player in the space started unloading size, and suddenly.. everyone remembered that big adoption can come with big exits too 💰️ 

BlackRock reportedly sold Bitcoin every single day over the past week, with the total reaching around ~$1.8 billion by the end of the week. According to data from Arkham that makes it BlackRock’s biggest Bitcoin ETF outflow since November 2025 💸 

The wider ETF market was not exactly glowing either. Combined outflows reportedly reached ~$2 billion for the week, with BlackRock carrying the largest share.

  • This all landed during a rough stretch for the market. BTC had slipped into the mid $70,000s, sentiment turned bad and investors started pulling capital instead of pretending every red week is just “healthy consolidation” 🤒 

  • When ETFs are pulling money in, the market treats them like adult supervision finally arrived. When flows reverse, the same structure suddenly reminds everyone that institutions are not emotionally attached to BTC or crypto. They do not care about cycle memes, laser eyes or someone’s 15th “supercycle” tweet.

The hyped up ETF adoption gave some crypto assets more access and a bigger pool of capital. It also made these crypto assets more exposed to traditional investor behavior. If macro goes bad ETF flows can turn from support into pressure very quickly 👇️ 

The ETF demand had become one of the market’s favorite comfort blankets. Every inflow week looked like proof that institutions were steadily absorbing supply and that we’ve found a permanent buyer class. A week like this ruins that bedtime story a little 🫠 

Ethereum is “outgrowing” its Foundation and old narratives

Ethereum is having one of those weeks where the chart looks not that great but the underlying story keeps getting stronger 👀 

On one side, with the aforementioned ETF exposure going down Ethereum ETFs have seen some net outflows in May as well, especially the BlackRock’s ETH ETF ownership has dropped, also Harvard has apparently stepped away from its Ethereum ETF position 😶 

On the other, we have to look at where some of the ETH is going.

  • Corporate Ethereum reserves have now climbed to around 7.33 million ETH, according to CoinGlass data. That puts roughly 6% of Ethereum’s total supply on corporate balance sheets, with strategic reserves valued around $16 billion 💵 

  • Two fresh whale wallets also pulled around $125.91 million worth of ETH, while BitMine’s accumulation pattern was compared to earlier buying behavior tied to its Russell 3000 inclusion 👇️ 

So it looks like ETF exposure is getting trimmed while direct corporate accumulation keeps building. Some players are stepping away from packaged exposure, while others are holding ETH like a treasury asset.

  • ETF flows can turn fast because they are easy to enter and easy to exit. Corporate reserves are usually harder to dump. A company putting ETH on its balance sheet is making a different kind of bet than a fund manager adjusting exposure after a bad month 🪙 

In the meantime, Vitalik Buterin has been making a related point from another angle: Ethereum itself should not revolve around the Ethereum Foundation.

He said the EF is moving toward being a smaller and more focused organization, not the center of Ethereum. More like one node with a defined purpose:

Vitalik also pointed out that the EF controls only around 0.16% of the total ETH supply, far less than some blockchain foundations that reportedly sit on much larger shares of their own networks.

  • His message was basically that Ethereum should keep leaning into longevity, censorship resistance, privacy, security and open development rather than chasing every metric just because another chain made a prettier dashboard.

  • The foundation has also been dealing with visible exits, including names like Tim Beiko, Barnabé Monnot, Josh Stark and Tomasz Stańczak. Naturally, people started reading that as drama, because in crypto there’s usually a war attached to literally anything 🤐 

ETH investor Ryan Berckmans argued the exits were more about strategy, leadership changes and internal restructuring than lost faith in Ethereum itself.

Ethereum seems to be in a strange but important phase. Corporations are holding more ETH, ETF demand is shaky but the Foundation is trying to become less centralized to address some of the concerns and the network keeps pushing toward a model where no single group gets to be the main character 🫡 

Google sold the top search spot to a fake Uniswap wallet drainer (again)

  • This week, people searching for Uniswap on Google were served a blatantly fake “sponsored result” sitting right where normal users are trained to click first. Not buried on page nine, right there at the top, dressed up as the thing people were looking for 🤡 

  • The fake ad impersonated Uniswap, complete with fake UI and pushed users toward a wallet duster contract and, according to onchain analysts, drained funds from multiple wallets. The attackers were reportedly holding at least $400,000 🪙 

Stacy Muur from Green Dots called it out on X, saying two scammers had already stolen around $400,000 through a phishing ad on Google 👇️ 

Users search for Uniswap, Google hands them a paid fake. Very elegant system. Truly peak internet civilization.

The best part, if you can call it that, is that the fake page was hosted through Google Sites. So Google did not just sell the ad placement. Its own product helped host the fake destination. A phishing campaign managed to walk through the front door, rent the lobby and put up a little sign saying “Official entrance here” 🤪 

That’s why it’s slightly absurd.

Google is perfectly happy to be nosy when ordinary users are involved. It will scan, flag, warn, rank, filter, recommend, prompt, detect “suspicious activity” when you’re doing your legitimite work, nag you about account recovery, bury and shadowban legit sites it thinks are not fit for its political agenda, label emails, analyze attachments and your cloud files and force half the web prove they are not robots by selecting traffic lights:

But a fake Uniswap ad sitting at the top of search, leading to a phishing setup, apparently slipped through the grand machinery of algorithmic supervision. Amazing how the all seeing empire suddenly becomes a sleepy raccoon when scam money enters the room 😵 

Many analysts have previously said fake Google ads are a common source of crypto phishing attacks. These ads imitate major crypto protocols, outbid real projects and get placed in the sponsored section above organic results 👇️ 

Sure, users should double check. Everyone in crypto knows the drill. Bookmark official sites, use trusted links, check X accounts, compare domains and treat sponsored results like a stranger offering free pills in a parking lot.

But pretending this is only a user education problem is lazy. Google has allowed the trap to be placed where trust is supposed to be. Search results are not random graffiti on a bathroom wall. You assume the top result has passed at least some basic test 💀 

Crypto already has enough traps built into it. Users should not have to survive Google Search too 🤦‍♂️ 

“The golden trio of onchain safety” according to Marc Zeller:

“Why Ethereum is materially mispriced, my fair value, and the framework people keep missing” by Tom Dunleavy

Major areas where the fin system needs an adjustment, according to Brian Armstrong:

MEMES

That's all for now, frens.

We'll meet in a week! And remember, the market conditions are temporary, but our commitment to building a better Web3 is here to stay. Thanks for joining us, and we look forward to seeing you back next week. Cheers!

Yours, The 🔥 Team

Brought to you by Ambire: The Only Web3 Wallet That You’ll Need!