When you’ve done an impossible thing, or a few, things stop seeming so impossible.
Do impossible things. It’s really quite fun.
007248
borski✓
@borski
Lead of @GoogleDeepmind Agentic Red Team | hacker | founder | cybersecurity expert | angel investor | singer/musician | UX enthusiast. All my views are my own.
602 Following870 Followers5.7K Posts
- borskiRT @local0ptimist: i had a chance to try Argon before joining deepmind but after using Carbon internally I can tell you that this team is absolutely cooking and we’re only just getting started. Carbon is writing all of my code and I can trust it to run all night, for multiple days, making intelligent decisions autonomously. there is still lots of work to do so we’re excited to get this in your hands and hear your feedback. we can’t build AI for everyone without the community 🖤Sundar Pichai @sundarpichaiI did say we'd be releasing our models periodically...introducing Gemini 4 Carbon, our next frontier model. Following Argon's release just two weeks ago, Carbon shows the pace at which we’re pushing the frontier, it ranks #1 on the recently updated v5.0 of @ArtificialAnlys Intelligence Index at 59. It also leads across key cybersecurity evaluations scoring 88.2% on SEC-bench Pro for finding vulnerabilities, 54.8% on ExploitGym for exploitation, and 71.0% on CWE-bench for patching. We’re seeing significant improvements in long-horizon tasks even compared to Argon, and we’ve strengthened frontier safeguards. As we did with Argon, we’re going to roll it out gradually + responsibly. Some trusted cyber defenders will get access starting this week. We'll be expanding it to more partners through our Fairwind Program, ahead of wider availability soon, so hang tight!06603
I did say we'd be releasing our models periodically...introducing Gemini 4 Carbon, our next frontier model. Following Argon's release just two weeks ago, Carbon shows the pace at which we’re pushing the frontier, it ranks #1 on the recently updated v5.0 of @ArtificialAnlys Intelligence Index at 59. It also leads across key cybersecurity evaluations scoring 88.2% on SEC-bench Pro for finding vulnerabilities, 54.8% on ExploitGym for exploitation, and 71.0% on CWE-bench for patching. We’re seeing significant improvements in long-horizon tasks even compared to Argon, and we’ve strengthened frontier safeguards. As we did with Argon, we’re going to roll it out gradually + responsibly. Some trusted cyber defenders will get access starting this week. We'll be expanding it to more partners through our Fairwind Program, ahead of wider availability soon, so hang tight!
1.9K2.5K31.2K5.4M- Carbon… really is solid. 😉 I’ll be here all weekSundar Pichai @sundarpichaiI did say we'd be releasing our models periodically...introducing Gemini 4 Carbon, our next frontier model. Following Argon's release just two weeks ago, Carbon shows the pace at which we’re pushing the frontier, it ranks #1 on the recently updated v5.0 of @ArtificialAnlys Intelligence Index at 59. It also leads across key cybersecurity evaluations scoring 88.2% on SEC-bench Pro for finding vulnerabilities, 54.8% on ExploitGym for exploitation, and 71.0% on CWE-bench for patching. We’re seeing significant improvements in long-horizon tasks even compared to Argon, and we’ve strengthened frontier safeguards. As we did with Argon, we’re going to roll it out gradually + responsibly. Some trusted cyber defenders will get access starting this week. We'll be expanding it to more partners through our Fairwind Program, ahead of wider availability soon, so hang tight!002163
- borskiRT @demishassabis: ⚡Iterating at lightning speed! Gemini 4 Carbon is SOTA across key benchmarks and combines frontier capabilities with deep safety & security research.036203
- Keep writing; it keeps you thinking.Paul Graham @paulgWe're about to enter a period where not reading or writing becomes as ubiquitous as smoking once was. We can hope people will eventually realize this is bad for them, as they did with smoking. But this is far from certain; the cost may not be as obvious.00047
We're about to enter a period where not reading or writing becomes as ubiquitous as smoking once was. We can hope people will eventually realize this is bad for them, as they did with smoking. But this is far from certain; the cost may not be as obvious.6381.3K14.3K1.1M- The more I spend time in the “AI industry,” the more I realize that security practitioners were “outside” and are now slowly working their way inside. For some reason, AI labs started by essentially pretending we’ve never learned about or spent time thinking about security. 1/21111.1K
- Posts @borskiThere are lots of new problems in securing AI. Lots. But there are also lots of old problems. We know how to solve many of them. But if all security problems are “new” to you, on average, please talk to some “old hat” security folks. Hire them. Teach them AI. It’s worth it. 5/10154
- Posts @borskiThere is nothing at all wrong with either “hacker.” But they are different. And too many people don’t understand (or care to understand) the difference. Is this a 🌶️ take? Dno. 6/EOM10145
- This is well said. And true. I have no idea what the future looks like. Neither does anyone else, and those who claim to are fortune-telling. We make bets. That is gambling, albeit with data and not random chance. I’m only very confident about the future being unpredictable.Zack Korman @ZackKorman@halvarflake The people who claim they know what’s coming unlike everyone else always seem the most surprised by every new thing they read.002105
Posts @halvarflake@halvarflake The people who claim they know what’s coming unlike everyone else always seem the most surprised by every new thing they read.56642.5K
- Literally went through the same experience. Getting acquired was, far and away, the most life-changing experience I’ve had. Most of those changes were good. But all of those changes were hard. And being hard doesn’t mean it’s whining; the “identity crisis” post-sale is real.Camilo Acosta @CamiloBAcostaI was a founder and CEO for 12 years before my company was acquired by Meta. Overnight, I went from running my own company to being one of tens of thousands of employees. I remember getting my employee badge, walking around campus with my backpack, and feeling like a college kid on his first day of school. People kept telling me I landed a dream job. And for a lot of people, product at Meta really is the dream! But it was never mine. I didn't even know how to introduce myself anymore. Was I a founder? Former founder? Product manager at Meta aka just another one of Zuck's lackeys? I've seen plenty of exited founders go through something similar. They can make incredible amounts of money but once the excitement wears off they feel completely lost. Their core identity has been blown up. If you're thinking about selling your company, talk to a few founders who've already done it. They can prepare you for the transition in a way your investors and lawyers can't.0121.1K
- I was a founder and CEO for 12 years before my company was acquired by Meta. Overnight, I went from running my own company to being one of tens of thousands of employees. I remember getting my employee badge, walking around campus with my backpack, and feeling like a college kid on his first day of school. People kept telling me I landed a dream job. And for a lot of people, product at Meta really is the dream! But it was never mine. I didn't even know how to introduce myself anymore. Was I a founder? Former founder? Product manager at Meta aka just another one of Zuck's lackeys? I've seen plenty of exited founders go through something similar. They can make incredible amounts of money but once the excitement wears off they feel completely lost. Their core identity has been blown up. If you're thinking about selling your company, talk to a few founders who've already done it. They can prepare you for the transition in a way your investors and lawyers can't.
2129110.7K - YupMatt Slotnick @matt_slotnickmaybe i'm taking crazy pills but winning big in AI, like winning really really big in AI, requires two things. it requires scale, and it requires vertical integration from workload through components of the data center... and Google has been preparing for that war for decades001145
maybe i'm taking crazy pills but winning big in AI, like winning really really big in AI, requires two things. it requires scale, and it requires vertical integration from workload through components of the data center... and Google has been preparing for that war for decades56797.7K- Show more posts
