Melak Wubshet π retweeted
Binance just published how $6.34 billion walked out the door.
June: an engineer searches for a popular AI coding tool, clicks a search result, and installs a tampered homebrew. Binance catches it, replaces the laptop, rotates the credentials. the attackers now know who works on the wallet team.
September: another engineer gets a job offer on a real job-hunting site. first-round interview with βHRβ over live video. βHRβ shares their social media, personal photos and videos included, so he trusts them. second round is with the βCEOβ of a company that doesnβt exist. at the end of the call heβs asked to download a take-home βassignmentβ onto his work laptop.
the assignment quietly swaps out pyenv for a malicious copy. No antivirus flags it.
october 20: the attacker uses binanceβs own internal AI tool to get into two more engineersβ machines. One of them has a container deployment token sitting on his laptop.
october 29: container deployed. Signing service reached. $6.34 billion moved.
Binance shut off withdrawals. the money kept moving anyway, because the attacker had already given their own smart contract permission over a hot wallet, on-chain. binanceβs off switch didnβt reach that far.
then they tried to revoke the permission. the wallet uses MPC, so thereβs no single private key to sign the revoke with. they had to build a revoke procedure on the spot while funds were still leaving.
last stolen ETH went through at 23:56 UTC. almost four hours after the first transfer.
Phew!
Melak Wubshet π retweeted
Vibe coding isnβt the problem. Your taste is.
I built 100 apps in 30 days. 2,000 people use them. They live in an app store I built myself.
Not one of them looks vibe coded.
Same AI you have. Different standards.
The UX for C02 calculator app, at the science museum should get attention and rebuilt.
p.s. Visiting Addis with my boss from HTM Technologies.
#AddisAbaba #Ethiopia
Cats are the most adorable and crazy creatures; Nest js too! π
@nestframework
#nestjs #backend #developers
Melak Wubshet π retweeted
Hosting Zero to Agent in Addis Ababa on May 2.
Global Build Week. Ship a real agent with @v0 + @vercel.
Free, $6k+ prizes up for grabs.
Donβt miss out secure your spot now :) luma.com/f9yxgtri
#ZerotoAgent
Melak Wubshet π retweeted
African Cities Innovation Fund β Grants Up to $75,000 for African Innovators.
Are you working on solutions to urban challenges in Sub-Saharan Africa? The #African Cities #Innovation Fund is supporting collaborative innovations that build smart, healthy, equitable, and resilient cities.
β¨ Key Benefits
1. Grant funding of up to $75,000
2. Support for collaborative, scale-ready urban innovations
3. Focus on climate resilience, digital equity, youth employment, and healthy urban spaces
4. Opportunity to pilot impactful solutions for low-income urban communities
π
Applications open: Early 2026
π Learn more & apply:β¨
lnkd.in/gXeTbzjZ
#AfricanOpportunities #GrantsForAfrica #UrbanInnovation #AfricaFunding #OpportunitiesForYouth
Advanced Machine Intelligence (AMI) is building a new breed of AI systems that understand the world, have persistent memory, can reason and plan, and are controllable and safe.
Weβve raised a $1.03B (~β¬890M) round from global investors who believe in our vision of universally intelligent systems centered on world models. This round is co-led by Cathay Innovation, Greycroft, Hiro Capital, HV Capital, and Bezos Expeditions, along with other investors and angels across the world.
We are a growing team of researchers and builders, operating in Paris, New York, Montreal and Singapore from day one.
Read more: amilabs.xyz/
AMI - Real world. Real intelligence.
My feed flooded with "one-shot" landing pages vibe-coded with #Gemini3 π.
Thread: π§΅
AI agents must be resilient, not just smart!
Execution stage in agentic workflow is the hardest stage for agents to achieve successfully. It could be due to tool call error (the most common one), output parsing fails, side-effect mismatch or env't changed mid-loop.
Top Systems have common "Execution Trick" to mitigate these failures. For example:
LangGraph - Roll back on failed execution, retry with new plan
OpenAI Swarm - Human-in-the-loop on execution failure
Semantic Kernel - Automatic function calling with execution setting
CrewAI - Max iterations to prevent loops or timeouts.
TL;DR:
Never trust execution! turn a 70% failure risk into 99.9% reliability using -> Validate-> Atomic write -> Verify -> Fallback
#AI #AIAgent #AgenticAI
Anticipating the surge of consumer and commercial robots by 2026. The real question is, how will we adapt to the unprecedented level of automation in the workforce and the scale of job displacement that's coming? #AI #Automation #Robotics
And this one is scary: xitter.gq/UnitreeRobotics/statusβ¦






















