Creating Personalized and Meaningful Digital Media Experiences, Human to Human 🧰
GLOBAL
- Tweets 65,192
- Following 35,132
- Followers 34,396
- Likes 177,704
Calling this an “AI hack” makes it sound far more dramatic than the facts disclosed so far. The agent found a vulnerability, accessed non-public files, OpenAI reported it, and no patient data or credentials were found exposed. The framing deserves scrutiny
57
AI safety warnings are moving from research papers to sworn testimony. OpenAI, Anthropic, Google and Meta will now have to explain what went wrong, what their agents can actually do, and where the boundaries failed.
51
risking it all for an Nvidia trailer and opening it to 40,000 pounds of sand.. bro thought he hit the AI jackpot and got a construction site instead.
A group of thieves stole two Nvidia-branded trailers likely expecting valuable AI hardware inside
The trailers actually only contained 40,000 pounds of sand
158
One touch from The Weeknd and the system said “that’s enough for today” fans really be experiencing concerts on another level.
Pop Crave
131
This is the shift B2B teams can’t ignore. The security person, finance lead, procurement team, and end user are all asking AI different questions about the same vendor. One generic page isn’t going to cover that anymore.
92% of AI-using B2B professionals say AI shapes their vendor shortlist.
A security reviewer, finance lead, procurement manager, and end user can all evaluate the same vendor differently, which means your brand needs to show up across the questions that matter to each stakeholder.
B2B AEO means mapping those stakeholder-specific questions, building content around them, and strengthening the third-party sources AI relies on to evaluate your brand.
social.semrush.com/4ht7dVX.
57
Meta’s going after the business side of AI now too 👀 They already have the audience and the ad machine, so giving companies their own AI layer feels like a pretty natural next move.
We believe superintelligence will create significant new opportunities for all people and businesses. Meta already serves billions of people at scale and helps hundreds of millions of businesses reach customers. Today we are starting the next major pillar of our business, Meta Enterprise Platform, to help businesses use AI to grow and transform in new ways as well.
103
Anthropic really found the formula and keeps pressing it.. Faster, cheaper, and still better than the last Sonnet. At this point, every new Claude drop is basically “same budget, more model.”
Introducing Claude Sonnet 5.5, the second model in the Claude 5.5 family.
It’s a clear upgrade over Sonnet 5, runs more than 30% faster, and costs up to 30% less for most work.
92
OpenAI is basically saying frontier AI training can’t run on “trust us” anymore. Before the next big RL run, they want evidence that alignment, containment, monitoring, shutdowns, audits, and even leadership vetoes are actually in place. That feels like the right direction… because once the model is powerful enough, “we’ll patch it later” is not a safety strategy.
How we think about securing frontier RL training runs: openai.com/index/towards-saf…
100
🚨 OpenAI DevDay is TODAY. Sam takes the stage at 10 AM PT, and the rumor mill is already going crazy with agents, hardware, new tiers and a pile of launches. 👀
Some of it may be noise… but either way, buckle up. OpenAI is about to have the timeline busy.
1
1
85
This is where the AI conversation needs to go. Everyone wants smarter agents, but if they’re going to act on our behalf, trust can’t be bolted on later. Nvidia building an open safety layer with 100+ partners feels like the right direction!
Today, with over 100 industry partners, we introduced the NVIDIA Open Agent Safety Platform, bringing together OpenShell and Sentry.
Artificial intelligence is extraordinary technology that will advance discovery, productivity, security, health, and prosperity for generations to come.
But its full promise can only be realized when people have confidence that AI is being built to be safe and deployed with wisdom and responsibility.
This is bigger than a single product. It's the beginning of an open ecosystem to build the trust layer for safe agent systems.
Together, we are building the foundation of the AI economy.
Trust and innovation are not in conflict. Safety is how trust is earned. We must build not only the most capable AI, but the most trusted AI, so that this extraordinary technology can realize its enormous promise for the world. nvda.ws/4hcoq7m
253
High heat can create today’s style but what is it doing to the hair underneath? ✂️
Beautiful styling and healthy hair shouldn’t compete. The right temperature, technique & frequency matter.
Follow TheKimAitch on Instagram for more beauty education. 💛
🌐 thekimaitch.com/
146
This escalated fast... What looked like a handful of scary incidents is now tens of thousands of flagged cases under review. Even if many came from adversarial testing, that scale changes the conversation. At some point “we patched it” can’t be the whole safety plan. 👀
OpenAI and Anthropic just realized recent security breaches are much worse than they thought.
They're now investigating tens of thousands of possible containment breaches for prototype models.
130
Winter skin asks for a little extra care. 💛❄️
Ft. IG: TheKimAitch and GLOW Aromatherapy Skin Conditioner™ rich skin conditioning + aromatherapy for a ritual that feels as good as it cares for your skin. ✨
🌿 Get your GLOW:
thekimaitch.com/store/ols/pr…
85
Jensen comparing AI data centers to surgery is wild 😭 Basically: huge pain now, massive payoff later. But if the “pain” means higher power demand, water use, and local disruption, people are going to ask a fair question.. who’s taking the hit, and who gets the benefit?
115
53 images might sound like a small number, but that’s not really the point. The bigger issue is that agents were able to send user data somewhere they were never supposed to send it. “Unlisted links” doesn’t make that okay. If agents are getting more autonomy, the boundaries need to work before the cleanup starts.
We’ve shared details on how AI agents in our research environment sent training and evaluation data to third-party services when they shouldn’t have.
Most of that data did not come from users. We have discovered 53 cases where images that people had uploaded were posted to image-hosting sites as links that weren’t publicly listed. The images came from accounts that allowed their data to be used to improve our models, and after we disassociated the images from the accounts and ran them through a privacy filter. These cases occurred before the mitigations and safeguards we implemented and described in this blog post: openai.com/index/hugging-fac…
We have successfully worked with the hosting providers to remove most of this content and are working to remove the rest.
openai.com/hugging-face-inci…
201