Konstantinos Sevdinoglou

Something is brewing.

🇬🇷 Alleged customer and loyalty-program dataset linked to Greek retailer “kotsovolos.gr” has surfaced on underground forums. According to the threat actor’s claims, the dataset allegedly contains approximately 476,000 records associated with: • Customer profiles • Orders and purchase activity • Loyalty program information • Support ticket interactions The post advertises access to structured datasets containing: • Full names • Email addresses • Phone numbers • Billing/shipping addresses • Birth dates • Marketing preferences • Loyalty status • Total spending history • Preferred currencies/languages • Customer segmentation metadata • Support ticket content If authentic, this type of dataset is significantly more valuable than simple credential leaks because it enables: • Advanced social engineering • Targeted phishing • Fraud profiling • Identity correlation • Behavioral analysis • Loyalty/rewards abuse • Financial scam personalization Retail and e-commerce ecosystems remain among the most consistently targeted sectors globally because they centralize: • Payment-related metadata • Consumer identity information • Purchase behavior analytics • Marketing datasets • Support systems • Third-party logistics integrations One important detail in this alleged leak is the inclusion of: • Customer support tickets • Purchase history • Marketing preference data These fields dramatically increase the operational value of stolen datasets for threat actors because they allow attackers to craft highly convincing impersonation campaigns. For example: • Fake delivery notifications • Loyalty reward scams • Refund fraud • Invoice impersonation • Customer support phishing • Warranty-extension scams become substantially more effective when attackers possess historical transaction context. Another increasingly important trend is the monetization of “consumer intelligence datasets.” Modern underground markets no longer value only: • Passwords or • Credit cards. Instead, attackers increasingly seek: • Behavioral profiles • Spending patterns • Customer segmentation data • Marketing metadata • Loyalty ecosystems • CRM exports because these datasets support: • Fraud operations • AI-assisted phishing • Synthetic identity creation • Large-scale social engineering campaigns European organizations face additional pressure because GDPR exposure from these incidents can become severe when: • Purchase behavior • Consumer profiling • Support communications • Contact metadata • Marketing consent records are involved. Retail organizations should continuously review: • CRM security posture • Third-party SaaS integrations • Loyalty platform exposure • Customer support tooling • API security • Cloud storage exposure • Data retention policies • Marketing platform access controls • Privileged account monitoring because attackers increasingly target: • Misconfigured APIs • SaaS integrations • Support systems • Marketing automation tools • Third-party plugins rather than the primary storefront infrastructure itself. The broader European retail sector continues to experience increasing pressure from: • Credential theft groups • Initial access brokers • Infostealer malware campaigns • Supply-chain compromises • Cloud misconfigurations • Ransomware operators especially where customer ecosystems contain high-quality behavioral and transactional data. At this stage, the underground forum claims remain unverified and should be treated cautiously until independently validated. #DDW #Greece #CyberSecurity #DarkWeb #DataLeak #RetailSecurity #EcommerceSecurity #ThreatIntelligence #GDPR #ConsumerData #LoyaltyProgram #FraudPrevention

@speedrun here it comes!

Is it me or Codex is eating up usage even on medium for you as well?

@LinusEkenstam You’re definitely cooking!

Could not find a 22 foot countertop, so I made one in micro-cement. With an anodized aluminum lip. Also decided to change the drawer fronts from light ash to this darker wood. The countertop is also deeper than normal, 73cm or 29 inches. Cooking channel incoming 🍓🥑🍳🍖

@Hostinger Already have 2 of them in use. Good stuff

Something big just landed for devs 💥 A new way to deploy Node.js apps is here — built for speed and serious security. ⚡️ 1-click deploy from GitHub 🛡️ Fully managed servers & security 💸 $3.99/mo flat. No usage-based fees Start deploying on Hostinger: hstn.gr/6L

@DataChaz only good tell is the size of the ID card which is wrong

this is where the problems really start

@thsottiaux Also EU when? 🇬🇷

@thsottiaux Playwright is playing up when I try to test more that one repo at the same time. Should I use docker for each instance instead? I think I’m wasting a lot of resources every time and even gets stuck on a single repo often.

Hello builders. What are we getting wrong with Codex, what can we improve?

On a personal level, I also have a one year old in the house. I would completely annihilate anyone who would try to harm my daughter. Since I don’t know you personally, I can only say that it’s utter nonsense to hunt one of the people responsible for bringing ChatGPT into the world and changing the trajectory of history. Whatever that turns out to be. Hope you feel better soon and take care of family first.

I wrote this early this morning and I wasn't sure if I would actually publish it, but here it is: blog.samaltman.com/2279512

@ChatGPTapp purchase parity and dollar to euro or uk pound conversion should be handled better. The new pro sub is $100 in the US but €103 in Europe and Greece. Dollar to Euro conversion is not 1:1, costs are not the same and income especially is not the same everywhere. There are few companies and platforms that do this well. I know you can do it too. There should be a call for all companies worldwide to do better. And yes there are VPNs etc. however people should not be forced to do that.

The biggest issue I see right now is: who has access to these models that we don’t even know of? Don’t forget that these are plain people (Dario, Sama, other country’s nationals like China) that actually control the release and use of these models. Since these are products that affect the planet and the very near future of our civilisation, shouldn’t there be some oversight from independent bodies? I also couldn’t sleep last night.

i'm on vacation with my family. i read about mythos and couldn't relax the rest of the day. i am completely stunned. i already have a severe case of ai psychosis. i dont know what to call this now. i'm up late right now (late for me). i can't stop reading about anthropic's new model that they can't even release publicly because it's so good. this feels different. words like "frightening" and "uneasy" and "scary" are being throw around by the anthropic team. i feel all of those things. i knew this moment was coming. i didn't know it'd be so soon. i'm generally optimistic. i don't feel as optimistic today. i was shell-shocked most of the day. my mind was stuck on it. i kept looking around at people enjoying their vacations with their families and...i just felt weird. like i had been told aliens are real, they're coming, and soon...and no one else knows. it's true though, practically no one knows what's happening in AI right now. where does this go from here? how quickly? is software solved? is all software vulnerable now? am i even asking the right questions? what about anthropic? this is an enormous amount of power for one company, one man (dario), to have. i've said this before but now it's more real than ever: can any company catch up to anthropic? opus likely helped build mythos, mythos will help build the next model after that. recursive self improvement is here. the "intelligence explosion" as leopold aschenbrenner put it, is here. i knew the frontier labs were racing towards ASI. i knew it. but i didn't fully grasp what it meant. the first company to reach it wins. period. full stop. nothing else matters. dario knew that and his bet on coding was right. on the one hand, imaging all science, math, coding, climate problems being solved. imagine cancer being cured. imagine going to the stars. on the other hand - imagine concentration of power, political and economic change happening so fast, society can't adapt. how do we go on like things are the same?

Link to the document: www-cdn.anthropic.com/8b8380204f7467…

Anthropic just published a 244-page system card for Claude Mythos Preview, and the real story is not “they made a better AI model.” It is that they appear to have built a model powerful enough that they chose not to release it to the public. According to Anthropic, Mythos is their most capable model yet. It shows a major jump in reasoning, software engineering, computer use, research, and knowledge work. But the part that should make people pause is cybersecurity. Anthropic says the model demonstrated such strong cyber capability that they limited access to a small group of partners for defensive use instead of making it generally available. That is not a normal launch decision. This is the part most people will miss: Anthropic also says Mythos is their best-aligned model yet by most of their measures. So the message is not, “the model is wild and uncontrollable.” The message is much stranger, and much more serious: the model is more aligned, more capable, and still risky enough that they decided broad access was a bad idea. That is the frontier now. The old question was whether these systems can become more useful. The new question is what happens when they become so useful, especially in dual-use domains like cyber, that even the labs building them hesitate to put them in everyone’s hands. Anthropic’s broader conclusion is basically this: catastrophic risk is still low overall, but the margin for error is shrinking, some evaluations are saturating, subjective judgment is becoming more important, and rare failures matter more when the system is this capable. That should concern anyone paying attention. Because once you get to this level, safety is no longer just about average behavior. It is about tail risk. A highly capable model does not need to fail often to create a serious problem. One bad edge case can matter more than a thousand normal interactions. To me, this system card reads less like a product document and more like a warning label for where frontier AI is heading next. Not panic. Not sci-fi. Not hype. A warning. We are moving out of the era of “look how smart this chatbot is.” We are entering the era of “what exactly have we built, who should be allowed to use it, and are the people building it even confident in their own measuring tools anymore?” When a frontier lab effectively says, “this model is too cyber-capable for normal release,” that is not background noise. That is the signal. x.com/AnthropicAI/st…

@SpatiallyMe Looks delicious!

I just launched a free App that lets you open Mac Apps…from your iPhone. And close them with a swipe. A much faster drag and drop. And a ton of more features in the pipeline. It’s called choclift and you really should try it out. Available for free in the App Store for iOS and macOS.

@WillManidis Thank you for this essay Will. I would love to read the document you passed around at some point.

@SnazzyLabs The X4 still fits the iPhone Air right?

Remember that really tiny e-reader I showed off a few months ago? Well they make a smaller one now that fits on the back of every iPhone with MagSafe.

@Gossip_Goblin Don’t stop creating. You’re past the AI tech tooling, into storytelling mastery. If I could invest right now, you would be getting stacks from me! Congrats on getting the recognition you deserve!

🥵

@venturetwins Count me in Justine!

Truly blown away by a new AI image model launching this week ✨ Finally, you can generate photos that actually look like you! It's so much better than everything I've tried - from LoRAs to NB Pro. Onboarding some early testers. DM or comment if you want access 👀

@LinusEkenstam Also…

The early majority getting a taste of Claude. We're going to be in one heck of a year. Accelleration guaranteed.

My “For You” page is broken. Every post now is an article or quotes one!