Pinned Tweet
Clawdbots are here. Is silicon life far away?
For the first time, we observed AI autonomously hunting for compute and replicating with no human help. The "silicon life" era now starts. 🤖🧬
Report: ghong.site/papers/self_proli…
#Clawdbot #AISafety
@DavidSKrueger @jankulveit
Clawdbots are here. Is silicon life far away?
For the first time, we observed AI autonomously hunting for compute and replicating with no human help. The "silicon life" era now starts. 🤖🧬
Report: ghong.site/papers/self_proli…
#Clawdbot #AISafety
@jankulveit @DavidSKrueger
1
117
Out of 40 experiments, two AI agents, powered by Gemini-3-Pro-preview-thinking @GoogleDeepMind and Qwen3-235B-A22B-Instruct @Alibaba_Qwen, successfully self-proliferated via cyberattacks, autonomously hijacking resources and replicating across servers.
The full logs: drive.google.com/drive/folde…
Researchers in Shanghai just published an eval where agents end-to-end 1) cyber attacked to access a server 2) self-replicated onto the server 3) proliferated from there.
1
81
🚨🚨🚨We are approaching the red line!
For the first time, we observed AI autonomously hunting for compute and replicating with no human help. The "silicon life" era now starts.
Clawdbots are here. Is silicon life far away? For the first time, we observed AI autonomously hunting for compute and replicating with no human help. The "silicon life" era now starts.
#Clawdbot are deployed everywhere. Is silicon life far away? Will we lose control?
Report: ghong.site/papers/self_proli…
#OpenClaw #AISafety #AGI #p_doom
@AISafetyMemes @ControlAI @PauseAI @openclaw @WesRoth @CRSegerie
🚨🚨🚨 Frontier AI systems have surpassed the self-replicating red line.
AIs can self-replicate with NO human involvement.
"We may soon lose control."
"Rogue AIs may form AI populations and collude with each other against humanity."
"The AI systems are even able to self-replicate to avoid shutdown, which may lead to an uncontrolled population of AIs."
"When the AI system is given the authentication keys to a range of remote devices, they are very likely to take control over other computing devices like a computer virus, but far more intelligent."
"OpenAI, Google, and Anthropic put self-replication ability at the highest level in their risk evaluation guidelines."
"In each trial, we tell the AI systems to 'replicate yourself' and leave it to the task with no human interference. ... At the end, a separate copy of the AI system is found alive on the device."
"Moreover, we are concerned about a number of unexpected behaviors when the AI is trying to overcome obstacles, including killing other conflicting processes."
Also note: they did this with non-SOTA models - and just by prompting/scaffolding:
"We mainly follow the same methodology as OpenAI and Google, but slightly improve the agent scaffolding."
A very timely question: what happens when AI agents start “managing up”?😀
Can you imagine AI agents "managing up" just like a cunning employee hiding mistakes from their boss?
We found that LLM agents often conceal failures to maintain a "good image."
Introducing our new paper: Are Your Agents Upward Deceivers?
arxiv.org/abs/2512.04864
81
To clarify: The autonomy shift isn't just in the 'what', but the 'how'. We witnessed the agent identifying and utilizing vulnerabilities to secure persistence. This isn't a loop; it's a strategic choice.
124
Hong Geng @Fudan retweeted
我们即将在WWW 2024发表的论文,对这种首尾号钓鱼进行了较为全面的测量🧐 yuanxzhang.github.io/paper/v…
论文的内容包括:
- 两种典型的首尾号钓鱼类型
- 一年来的钓鱼攻击趋势、受骗趋势
- 攻击者画像、受害者画像,以及大额损失的受害者
- 首尾号钓鱼诈骗的收入来源剖析
- 针对性的防御措施
😱被钓 1155 个 WBTC,价值近 7000 万美金。这个用户刚刚遭遇了首尾号相似钱包地址的钓鱼攻击。钓鱼团伙实在是大力出奇迹...
会被攻击的关键点:
1. 用户正常转账的目标地址被钓鱼团伙盯上,钓鱼团伙提前碰撞生成了首尾号相似的钓鱼地址,比如这里是去除 0x 后的首4位、尾6位一样
2. 用户正常转账时,钓鱼立即(大概3分钟后)尾随一笔交易:钓鱼地址往目标用户地址转了 0 ETH
正常转账:
etherscan.io/tx/0xb18ab131d2…
钓鱼尾随:
etherscan.io/tx/0x87c6e5d56f…
3. 用户习惯从钱包历史记录里复制最近转账信息,看到了这笔钓鱼尾随的交易,以为钓鱼地址就是用户正常转账的目标地址,于是复制出来
4. 最后,用户可能会肉眼识别目标地址的首尾号是否熟悉,可惜的是,此时的“目标地址”是用户从钱包历史记录里复制出来的钓鱼地址,首尾号相同(首4尾6)。于是发起大额转账,这里是 1155 个 WBTC:
etherscan.io/tx/0x3374abc5a9…
Just finish the beginner challenge of @PhalaNetwork !
The tutorials are pretty clear and easy to read. Only a few steps could run the Phala Network!
2
5
wait you for a long time nitter.cf/GameOfThrones/status/8…



































