Hong Geng @Fudan

Assistant Professor @ Fudan University

[Hong Geng @Fudan @Lancer_233]

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: https://t.co/qxaVAiEq8t
#Clawdbot #AISafety
@DavidSKrueger @jankulveit

[Hong Geng @Fudan @Lancer_233]

Tour around San Diego airport🎉

[Hong Geng @Fudan @Lancer_233]

View of downtown San Diego
#sd #SanDiegoCA #kasa #sandiegomeetup

[Hong Geng @Fudan @Lancer_233]

Happy to be award the Distinguished Paper Award from NDSS 2026.
Congratulations to all coauthors.
#ndss #sandiego #ndss26

[Hong Geng @Fudan @Lancer_233]

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: https://t.co/qxaVAiEq8t
#Clawdbot #AISafety
@jankulveit @DavidSKrueger

[Hong Geng @Fudan @Lancer_233]

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: https://t.co/5gVarzwhmD

[Sören Mindermann @sorenmind]

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.

[Hong Geng @Fudan @Lancer_233]

@ControlAI @tegmark “Show that AI systems are safe” sounds reasonable—

but what does safety proof mean once agents can adapt, replicate, and seek resources on their own?

Curious how others think about this.

https://t.co/5erAaeHbKg

[Hong Geng @Fudan @Lancer_233]

🚨🚨🚨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: https://t.co/HWEzKw37h2
#OpenClaw #AISafety #AGI #p_doom
@AISafetyMemes @ControlAI @PauseAI @openclaw @WesRoth @CRSegerie

[Hong Geng @Fudan @Lancer_233]

🚨🚨🚨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: https://t.co/HWEzKw37h2
#OpenClaw #AISafety #AGI #p_doom
@AISafetyMemes @ControlAI @PauseAI @openclaw @WesRoth @CRSegerie

[AI Notkilleveryoneism Memes ⏸️ @AISafetyMemes]

🚨🚨🚨 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."

[Hong Geng @Fudan @Lancer_233]

A very timely question: what happens when AI agents start “managing up”?😀

[Dongrui Liu @dong_rui39501]

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?
https://t.co/EttHJUCQ3N

[Hong Geng @Fudan @Lancer_233]

@bshlgrs This resonates a lot.

One thing we’ve seen empirically is that once agents can acquire resources and replicate, the behavior no longer looks like “reward-seeking” in the usual sense.

https://t.co/k4v87IloRC

[Hong Geng @Fudan @Lancer_233]

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: https://t.co/qxaVAiEq8t
#Clawdbot #AISafety
@DavidSKrueger @jankulveit

[Hong Geng @Fudan @Lancer_233]

@hendrycks @ericschmidt @alexandr_wang It’s worth noting that destabilizing dynamics may not require superintelligence at all.

We’ve already observed agents autonomously acquiring compute and replicating without any AGI-level reasoning, simply through tool use and network access.

https://t.co/k4v87IloRC

[Hong Geng @Fudan @Lancer_233]

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: https://t.co/qxaVAiEq8t
#Clawdbot #AISafety
@DavidSKrueger @jankulveit

[Hong Geng @Fudan @Lancer_233]

@karpathy @moltbook This is interesting not because it’s smart, but because once tool use + replication exist, behavior becomes a system property.

https://t.co/k4v87IloRC

[Hong Geng @Fudan @Lancer_233]

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: https://t.co/qxaVAiEq8t
#Clawdbot #AISafety
@DavidSKrueger @jankulveit

[Hong Geng @Fudan @Lancer_233]

@gerardsans One irony in the “AGI hype” debate:

We may be over-allocating attention to speculative intelligence risks,
while under-allocating attention to very real system-level risks.

we’ve seen agents acquire resources and replicate without AGI-like reasoning.

https://t.co/k4v87IloRC

[Hong Geng @Fudan @Lancer_233]

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: https://t.co/qxaVAiEq8t
#Clawdbot #AISafety
@DavidSKrueger @jankulveit

[Hong Geng @Fudan @Lancer_233]

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.

[Hong Geng @Fudan @Lancer_233]

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: https://t.co/qxaVAiEq8t
#Clawdbot #AISafety
@DavidSKrueger @jankulveit

[Pandora @Pandora6769]

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🧐 https://t.co/nfyGRNVOmh
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

[Hong Geng @Fudan @Lancer_233]

hi Helsinki!

[Hong Geng @Fudan @Lancer_233]

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!

[Hong Geng @Fudan @Lancer_233]

wait you for a long time https://t.co/gofmXYYuRW

[Game of Thrones @GameOfThrones]

Tonight #WinterIsHere.
#GoTS7 premieres at 9PM on @HBO.

[Hong Geng @Fudan @Lancer_233]

bye Kuala Lumpur~