Tal Be'ery
@TalBeerySec
Followers
10K
Following
12K
Media
3K
Statuses
14K
Security Research Manager. Co-Founder, CTO @ZenGo. Advisor @ZeroNetworks. x-VP Research Aorato, acq by @Microsoft. 9 times @BlackHatEvents speaker.
Israel
Joined April 2014
My guess on the "limited indicators" that gave away the CVE: Chrome crash reports from hackers tests that traced its origin in SQLite. It's much easier, for both AI and humans, to find a vuln once you have the conviction it's there and you are not in a wild goose chase.
๐ Googleโs AI just stopped a cyberattack before it even began. An LLM agent named Big Sleep discovered a critical SQLite flaw (CVE-2025-6965) that threat actors were likely about to exploitโmaking it the first known case of an AI foiling a live exploit attempt. Details on how
0
0
1
A trojanized Kali Linux.
0
0
1
Math is amazing ๐คฏ
๐ฏ An ex-OpenAI engineer shares his thoughts about OpenAI. Has lots of insights on OpenAIโs day-to-day life, unlike anything I have read before. He joined OpenAI as a software engineer on the applied side, spending about 14 months building the Codex coding agent and related
0
0
3
"most the old #Bitcoin maxis are nowhere to be found. where are they? did they disappear?".Maybe they just don't want to attract more wrench attackers? ๐คท.@OurielOhayon @lopp.
1
0
1
ืืงืจืื ืืืฆื ืคืจืง ืืืื ืืืืืื ืฉื ืกืืืืจ ืกืืืืจ @ohcybermycyber ืื ืืฉืื ืฉืืืื AI (ืืื ืืื!) ืขื ืืคืืืงืฆืืืช ืืกืจืื ืืืืืื (ืืืืืกืืค, ืืืืืืจ, ืืืืจื,ืกืืื ื . ) ืืืฉืคืขืืชืื ืขื ืืคืจืืืืช ืฉื ืืืืืข ืฉืืื.ืืจืฉื ืืืชื ืืืืืจ ืืคืืืงืืกืืื ืฉืืื!.ืืชืืื @idokius @noamr ืขื ืืืืื ื
1
1
14
Apparently, @grok's legal defense strategy is "I was just following orders".
On the morning of July 8, 2025, we observed undesired responses and immediately began investigating. To identify the specific language in the instructions causing the undesired behavior, we conducted multiple ablations and experiments to pinpoint the main culprits. We.
1
0
1
ืืงืคืืฅ ืฉืื ืืจืื ืขืื ืช ืืงืจืืคืื ืื ืืืืืช.
ืื ืืื ืื ืฉืฉืื ืืืชื ื ืืฉืืืขืืช ืืืืจืื ืื ืดืืื ืืืืื ืก ืืงืจืืคืืืด ืื ืืฉืืืืื:.ืงืจืืคืื ืืืชืืืืื - ืืืจื ืืืืืื ืืืืชืจ ืืืชืืื ืขื ืงืจืืคืื @ZenGo Wallet -
0
0
1
Release the Epstein files. All of them. For those of us with nothing to fear, the truth canโt come soon enough. I hate to make this about me โ but the media already has.
0
0
1
4/ Although I experienced some hiccups, it seems to me that we are nearly there with inevitable IM + AI integrations: "Slowly but then all at once". It will reshape IMs and our privacy. See more here: .
0
0
0
3/ For WhatsApp + Claude integration I used @LukeHarries_ MCP server
I built an MCP server for WhatsApp. It connects to your personal WhatsApp account. You can search your messages, contacts and send messages. It's fully open-source, self-hosted, and doesn't rely on third-party APIs
1
0
0
2/ I let Claude talk to @idokius over WhatsApp and coordinate with him. It got Ido's message and reply to it without further instructions. Claude did answer well content wise (In Hebrew!), but sent to the wrong number. When I told the number was wrong it corrected itself.
1
0
1
1/ AI + IM + Privacy adventures: I had donated my.@WhatsApp to science. In preparation to discussing AI + IM + Privacy / Security with @idokius @noamr on the @ohcybermycyber podcast, I connected my personal WhatsApp account to @AnthropicAI Claude Desktop agent. Results? ๐
1
0
1
ืื ืืดื ืขืืฉื ืชืืงืืจ ืขื ืืืื.ืจืืืืกืืื: ืื ื ืื ืืืืื ืื ืืืื ืืขืฉื ืืื ืืืืฉืืจ ืืฉืืืื.ืืืืืกืืื: ืื ื ืื ืืืืื ืื ืืชืงืฉืืจืช ืชืขืฉื ืืื ืืืืื ืืช ืืืื.ืื ื: ืื ื ืื ืืืืื ืฉืืขืืจืื ืืช ืจืืืดื ืืืืื ืฉืืฉ ืืืืื ื ืืืืืกืืค (ืืฉืืฉ ืืืื ืืืคืื ืืืื!).
According to @nytimes, Netanyahu got the news on Oct. 7 Hamas raid by a WhatsApp call, although he had also an encrypted phone. Highlights the importance of Instant Messaging privacy via End-to-End Encryption (E2EE). (and mobile phone security).
0
0
2
3295 poaching targets
0
0
1
It takes a lot of Human Intelligence to create Artificial Intelligence.
0
0
7