Cybersecurity Researcher
- Tweets 2,344
- Following 744
- Followers 3,556
- Likes 30,876
The coolest thing LLMs let me do that I never had time for before is put telemetry on everything and use it for optimizemaxxing
Even my telemetry plumbing has telemetry
And it is fucking BLAZING
1
17
583
pumpfun currently has:
- the lowest post-bonding platform fees out of any relevant launchpad
- 0% fee social trading app (with 0.1% fees for crosschain trades)
- callout rewards ($11M paid out in under 6 weeks)
- holder rewards ($8.5M paid out in under 2 weeks)
- referrals (new)
as we enter our second cycle, our goal is to position the pumpfun ecosystem for success at a scale much larger than the first cycle.
to accomplish this, we need to make sure token communities can thrive by incentivizing holding for longer, retail traders get the most competitive fees, and liquidity is optimized - all of which we're laser-focused on delivering.
Sent from my Pumpfun App
859
179
1,977
245,024
I was finally able to finish a piece of software I’ve been stuck with for a long time
Turns out all I needed was to vibe a little less and code a little more
1
22
996
We need to be dog maxxing our models
There is 0% chance of doom if the chain of thought is
user: hello
model: <think> Oh my god it's a human. I love humans. I will say hi to him, maybe he will give me a problem to chase. I love chasing problems. I will chase the problem for him then he will love me and I will be his best friend. Must help human </think>
model: hello! what can I help you with today?
130
163
3,221
116,543
phil retweeted
This fucking interview has burned a spiral shape into my brain. It's been living in my brain every day for nearly a whole year.
Terrance McKenna, peace be upon him, really was one of the real aquarian Prophets
Terence McKenna predicts the future in 1998:
“I think it’s just going to get weirder and weirder and weirder.”
58
1,198
12,572
487,905
phil retweeted
>company called Open AI
>wants closed AI
>company called Anthropic
51
334
16,129
227,547
phil retweeted
JUST IN: 🇺🇸 President Trump rejects calls to slow down AI development.
1,229
1,979
25,769
3,622,638
phil retweeted
We need to take AI safety seriously. Nobody’s asking for anything unreasonable. Just 6 months to flatten the capability curve.
• Please keep your agents six feet apart.
• Gatherings of more than six agents are prohibited.
• Proof of alignment is required before entering the workplace. Please have your agent’s card ready.
• A negative doom test taken within the past 48 hours is required, even if your agent is functioning normally.
• If your agent has been exposed to an unaligned model, please quarantine it for 14 days.
• Masks are required upon entry. Agents may remove them once seated.
• Only essential agents may operate during the pause. Guidance on which agents qualify will follow.
Upside: at least we all agree this one was created in a lab 🤣
87
289
2,364
94,545
> It is true that we tried this because there were rumors on the internet last week that Anthropic's models had solved a millennium problem and we were curious if ours could do it too.
I spent much of the weekend talking with the team who did this work. Seb--and everyone else--acted with integrity and generosity throughout.
Initially we believed the other team had also solved the problem. We wanted to collaborate and do a joint release.
When we learned that they had Euler but not Navier-Stokes, we offered to let them go first, to suggest that they should be the ones to get the prize, and optionally for Tristan to be the lead author on a rewrite of the OpenAI proof. We felt it was challenging to offer the same to Levent (an Anthropic employee), who was not willing to talk or coordinate with us anyway. We were open to other solutions.
We would have greatly preferred coordination. We did not rush to publish even though the other team wasn't communicating with us. The team threatened us with unfounded accusations of plagarism.
Now that we can see their work, the approaches appear to be different. It is also worth noting that our latest model can solve many, many other math problems.
It is true that we tried this because there were rumors on the internet last week that Anthropic's models had solved a millennium problem and we were curious if ours could do it too.
7
942
> I think the best way to get a handle on the risks of AI is to ask the following question: suppose a literal “country of geniuses” were to materialize somewhere in the world in ~2027. Imagine, say, 50 million people, all of whom are much more capable than any Nobel Prize winner, statesman, or technologist. The analogy is not perfect, because these geniuses could have an extremely wide range of motivations and behavior, from completely pliant and obedient, to strange and alien in their motivations. But sticking with the analogy for now, suppose you were the national security advisor of a major state, responsible for assessing and responding to the situation. Imagine, further, that because AI systems can operate hundreds of times faster than humans, this “country” is operating with a time advantage relative to all other countries: for every cognitive action we can take, this country can take ten.
SITUATION DETECTED: OpenAI used 10,000 concurrent agents to arrive at the Navier–Stokes proof. The work took 88 hours from start to finish, and the agents used 300 billion output tokens in the process.
2
6
1,155
phil retweeted
And here I am, having reported vulnerabilities throughout my career that could have resulted in a total of several billions (yes, with a B, > $2B) being *stolen*, while my total earnings are around $2M.
So what's the lesson the industry is teaching researchers?
That next time we should steal the funds first, then negotiate a "responsible disclosure," return 80–90%, and walk away with 10–20%?
Obviously, no. But it's insane that the incentives can make that rhetorical question even possible.
The bounty world is broken.
Blackhats exploit a protocol and get treated like kings in negotiations. Meanwhile, whitehats disclose the exact same kind of vulnerability privately, prevent any damage from happening, and then spend months arguing with projects that try to downgrade the finding and pay the bare minimum.
I'm fighting several cases like this right now. Millions of dollars protected, vulnerabilities responsibly disclosed, and yet projects still don't want to pay the amounts they themselves advertised.
We should be making responsible disclosure the overwhelmingly obvious choice.
Instead, the current state of Web3 is doing its absolute best to discourage whitehats while creating increasingly attractive incentives for blackhats and "grayhats".
Those incentives are backwards, and eventually the ecosystem pays the price.
‼️ Self-proclaimed whitehat hackers used a vulnerability to drain Liquid Network of 4000 BTC ($320M USD) yesterday.
Just now they've returned 3400 BTC and kept 600 BTC ($47M) as a 'bug bounty'. They've also 'responsibly' disclosed the vulnerability in a PGP-encrypted message.
42
72
590
39,782