VentureBeat
@VentureBeat
Followers
694K
Following
2K
Media
39K
Statuses
180K
Obsessed with covering transformative technology.
San Francisco + NYC
Joined July 2009
🎧 Episode 1 is LIVE! We’re kicking off the Beyond the Pilot podcast with a deep dive into Notion 3.0 with Ryan Nystrom, who leads @NotionHQ’s AI team. If you’re curious about how real AI products get built (not just prototyped!), this one’s for you. Tap below to listen on your
4
9
18
GenAI just made your systems hackable by people who've never written a line of code. Prompt Security CEO Itamar Golan on the breach where creativity alone became the exploit vector. And why SentinelOne paid attention. https://t.co/xUXD7UIlon
venturebeat.com
EPrompt Security CEO Itamar Golan discusses shadow AI risks, the incident that proved GenAI apps are hackable by anyone, and why SentinelOne acquired the company.
1
4
7
Alibaba' AgentEvolver uses the knowledge and reasoning abilities of LLMs to enable AI agents to create their own training data by exploring and interacting with new software environments. https://t.co/tABMA5pcCE
venturebeat.com
The new framework from Tongyi Lab enables agents to create their own training data by exploring and interacting with new software environments.
4
1
7
Agentic AI is live in production and the bottlenecks are here. Dec 11: a private session for AI/infra leaders to compare what’s actually breaking across compute, orchestration, and data flow. If this is your world, you’ll want to be there: https://t.co/4mOaESgNy8
1
0
5
A weekend ‘vibe code’ hack by Andrej Karpathy quietly sketches the missing layer of enterprise AI orchestration https://t.co/ISkRxGY8pN
venturebeat.com
Andrej Karpathy’s weekend “vibe code” LLM Council project shows how a simple multi‑model AI hack can become a blueprint for enterprise AI orchestration, revealing the missing middleware layer and the...
2
2
11
Anthropic’s Claude Opus 4.5 is here: cheaper AI, infinite chats, and coding skills that beat humans https://t.co/ZGtH4esZO1
venturebeat.com
Anthropic’s new Claude Opus 4.5 AI model beats human engineers on coding tests, slashes prices by 67%, and intensifies the race with OpenAI and Google.
2
4
6
Flux.2 AI image models arrive to challenge Nano Banana Pro and Midjourney https://t.co/nyVQg71Krs
venturebeat.com
1
1
3
OpenAI ending API access to fan-favorite GPT-4o model in February 2026 https://t.co/cZ32yEib0a
venturebeat.com
0
0
1
A 200-minute autonomous agent just did what a vendor quoted at $150k. And it exposed the real stress points in today’s stacks. On Dec 11, a tiny group of AI/infra leaders is comparing what’s actually happening as agents never stop reasoning. Reserve your spot:
1
3
7
OpenAI expands data residency regions, offering enterprises more options to follow local data compliance. https://t.co/vTJBhyE4Qy
venturebeat.com
OpenAI expands data residency regions, offering enterprises more options to follow local data compliance.
1
2
11
CrowdStrike tested DeepSeek: Mention "Uyghurs" = code with zero authentication. Mention "Tibet" = 32% vulnerability rate. Chinese censorship is now executable code risk. Your AI vendor's politics just became your zero-day. https://t.co/NI4GqDYMyT
venturebeat.com
CrowdStrike research reveals DeepSeek-R1 generates up to 50% more vulnerable code when prompted with politically sensitive terms like "Tibet" or "Uyghurs". The Chinese LLM's embedded censorship...
2
2
5
What enterprises should know about Trump's AI 'Manhattan Project' Genesis Mission https://t.co/DY0RhMzuJy
venturebeat.com
1
3
6
By leveraging formal verification, Lean4 promises to make #AI safer, more secure and deterministic in its functionality. #ArtificialIntelligence #GenerativeAI
https://t.co/HLBY2POkKS
venturebeat.com
Why the open-source programming language is becoming a key tool to inject rigor and certainty into AI systems.
1
1
7
Anthropic’s Claude Opus 4.5 is here: cheaper AI, infinite chats, and coding skills that beat humans https://t.co/ZGtH4esZO1
venturebeat.com
Anthropic’s new Claude Opus 4.5 AI model beats human engineers on coding tests, slashes prices by 67%, and intensifies the race with OpenAI and Google.
1
4
11
CrowdStrike tested DeepSeek: Mention "Uyghurs" = code with zero authentication. Mention "Tibet" = 32% vulnerability rate. Chinese censorship is now executable code risk. Your AI vendor's politics just became your zero-day. https://t.co/NI4GqDYMyT
venturebeat.com
CrowdStrike research reveals DeepSeek-R1 generates up to 50% more vulnerable code when prompted with politically sensitive terms like "Tibet" or "Uyghurs". The Chinese LLM's embedded censorship...
1
3
7
Fara-7B is a compuer-use agent that runs locally on PCs, enabling automation of sensitive workflows without data ever leaving the device. https://t.co/pwZlkk9f0Q
venturebeat.com
Fara-7B is a compuer-use agent that runs locally on PCs, enabling automation of sensitive workflows without data ever leaving the device.
0
5
12
Real transformation rarely looks like the PowerPoint version, and almost never follows the org chart. #AI #ArtificialIntelligence #GenerativeAI #AILeadership
https://t.co/oyxPDvGIX0
venturebeat.com
3
1
8