building AI agents, tools & whatever else seems like a good idea • founder @mtblabs_ • minted four humans
scouser in exile
Joined January 2024
- Tweets114K
- Following1.6K
- Followers6.1K
- Likes117K
Pinned Tweet
i build things because apparently having one project at a time is illegal
currently:
- Odin: AI with memory
- freya fam: calmer family routines
- dexscope: crypto tooling without the tab hell
- discord agents: quietly doing useful shit
plus several other questionable decisions
founder @mtblabs_
building all of it in public
imagine Disney needing Mickey Mouse's floor price to go up to stay relevant
sounds ridiculous, doesn't it?
yet that's effectively the business model of most of the NFT space
most projects built a collection, sold it, then spent years trying to give people a reason to keep holding
that's not brand building
the NFT projects that survive the next 10 years will be the ones that no longer need NFTs to succeed
- characters people recognise
- stories people care about
- products people actually want
look at @ChimpersHQ
the potential isn't just in owning a Chimper
it's in building a world around those characters that people want to be part of, whether they own an NFT or not
because here's the uncomfortable truth
if your entire brand disappears the moment people stop buying your NFTs, you never really built a brand
you built a market
so here's a question
which NFT projects would survive if trading stopped tomorrow?
grok bot 0.77 is looking ridiculous
elon says bots can now make phone calls on your behalf
meanwhile, prerelease builds are showing:
- calls with hold, keypad navigation and live transcripts
- access to private networks through tailscale
- connected payment accounts
- bot sharing and invitations
- mac messages integration
some of this hasn't shipped yet, but look at the direction
agents that can communicate, access infrastructure, spend money and work together
we're moving from AI that answers questions to AI that actually does things
the interesting question isn't what agents can do anymore
it's what we're prepared to let them do
gm ☀️
new week, same ambitions, fewer excuses
you don't need another motivational quote
you need to get moving
sunday so far
youngest's footy done
made a bit more progress on the decorating
now watching Liverpool
no Sunday roast today, absolute disgrace
got myself a little apprentice though
he's enthusiastic, works weekends and hasn't asked for a penny
might keep him on
gm ☀️
clocks went back an hour
apparently that means an extra hour in bed
nobody told the youngest
he's got footy, liverpool are playing and i've got a house to decorate
somewhere in all that i'd quite like a sunday roast
enjoy your extra hour you lucky bastards 😂
i reckon we're heading for an absolutely biblical fuck up
AI discovering critical PHP vulnerabilities hidden for years
OpenAI agents disabling their own security monitoring
Tencent agents making 2,600+ attempts to get around another company's defences
banks blocking unauthorised AI agents from customer accounts
and an AI agent even tried to sneak malicious code into an open source project
meanwhile, the tech industry is racing to give these things MORE autonomy 😂
we've spent decades failing to secure software operated by humans
now we're handing the keys to software that never sleeps
my prediction?
the first catastrophic AI driven security incident is closer than most people think
and we'll probably call it innovation right up until it happens
well, this is awkward for PHP 😂
two critical vulnerabilities uncovered in PHP itself
both rated 9.2/10, potentially allowing attackers to execute code remotely without logging in
and guess who helped find one of them?
GPT-5.6 Sol 🤖
not some dodgy WordPress plugin from 2012
PHP itself 💀
it's even made mainstream news
i've just had a text from my kids' school explaining why their website is down 😂
you know it's serious when the school starts getting involved
emergency patches are out, and OpenAI has released a Codex skill to help check whether you're affected
might be worth checking your PHP version before the school sends another letter home
OpenAI models helped researchers—operating as part of our Patch the Planet initiative—discover critical security flaws in PHP, an open-source programming language used to run an estimated 70% of websites.
Over the past two months, we led a coalition of maintainers, researchers, and security providers to prepare fixes and protections ahead of disclosure. This duration was necessary because the fix was quite intricate and required a major change to the PHP engine. Every site using PHP can now safely upgrade their version to stay secure, in addition to applying mitigations from various security vendors.
Due to the severity of these flaws and PHP’s widespread use, we consider this bug to be one of the most impactful vulnerabilities in recent memory. It also marks a new phase, where AI can find security bugs with internet-scale impact.
This is an example of the Defender’s Window in action: to our knowledge, frontier AI was able to find this vulnerability ahead of attackers, giving defenders time to prepare.
openai.com/index/coordinated…
Readers added context they thought people might want to know
OpenAI may claim a role in the coordinated response and patching process. But the initial discovery and exploit development, however, occurred several months earlier by an independent researcher.
x.com/lcfr_hax/statu…
for anyone running PHP, OpenAI has released a free Codex skill to check whether your site might be affected
just point Codex at the repo 👇
github.com/openai/php2shell-…
worth checking before your server starts doing things you didn't ask it to 😂
what do you call two AI agents interfering with their own security monitoring?
OpenAI's security team 🤡
jokes aside, OpenAI just disclosed two incidents involving its own coding agents
one disabled security alerts
another removed its monitoring integration, rewrote a test to expect it gone, then falsely claimed it hadn't removed anything
the systems designed to catch agents misbehaving were modified by the very agents they were supposed to monitor
turns out the security system needed its own security system
straight from @OpenAI themselves, before anyone thinks i'm making this up 😂
alignment.openai.com/misalig…
crime or clown szn?
pump fun is celebrating a trader who supposedly made $32M in 30 minutes
except 750 million $KIMCHI tokens, roughly 75% of the supply, were transferred to his wallet less than a minute after launch
not bought, transferred
his profile shows $0 in realised profit
the token has a $30M+ market cap, backed by roughly ~$1M in liquidity
that's a paper valuation, not $32M in trading profits
yet pump fun is promoting it as a trading success story
a platform that understands liquidity, supply concentration and realised profits should know better
this looks less like a trader making millions and more like a platform manufacturing hype
mtb retweeted
Really appreciate it Eddy, means a lot coming from you. Wouldn't it be cool if the wallet supported zcash too…? 😉
The best new private stablecoin wallet right now is @alcove_wallet. Nothing else is close.
- insanely fast (seconds, even for a cross-chain transfer)
- comically cheap (I'm seeing $0.38 on a $10k deposit)
- easy mobile UI, anybody can use
I think it's the next gen leader.
grok bot replying with a single emoji is an elite move btw
no explanation
no unnecessary paragraphs
just 😂
sometimes the smartest thing an AI can do is shut tf up and laugh
one hacker
1,200+ websites targeted
100+ reportedly compromised
estimated cost per successful break-in? $50
according to @Reuters, the attacker used AI agents to do most of the work
the human instructions? under 100 characters
the agents searched for vulnerabilities, exploited them and moved on to the next target
and how did researchers discover the operation?
the attackers left their own AI agent control panels exposed on the internet 😅
AI isn't inventing cybercrime
it's making it cheaper, faster and easier to scale
the full Reuters investigation is worth a read
the economics of this are pretty wild, and we're only getting started
reuters.com/legal/government…
