Search the archive

Register today →

EVERYTHING OPENAI ANNOUNCED AT DEVDAY 2025

OpenAI is transforming ChatGPT from a chatbot into an operating system. The company launched Apps SDK so developers can build actual apps inside ChatGPT instead of just GPTs, meaning you can talk directly to Canva, Zillow, and Coursera without leaving the chat interface. OpenAI also released AgentKit for building agentic workflows with evaluation tools included.

NYSE OWNER ICE COMMITS UP TO $2B INVESTMENT IN BETTING PLATFORM POLYMARKET

The NYSE owner, Intercontinental Exchange (ICE), plans to invest up to $2 billion in the prediction marketplace Polymarket, valuing it at around $8 billion. This move seeks to provide legitimacy and expand Polymarket's reach, potentially aiding its return to the US market. ICE will also distribute Polymarket's event-driven data and collaborate on tokenization projects.

AMD STOCK SKYROCKETS 23% AS OPENAI LOOKS TO TAKE STAKE IN AI CHIPMAKER

OpenAI will deploy 6 gigawatts of AMD Instinct GPUs over multiple years, starting with 1 gigawatt in late 2026. If the deal is fully exercised, OpenAI would acquire 10% of AMD. Two weeks ago, OpenAI agreed to a deal with Nvidia for $100 billion, with OpenAI's data center spending commitments now totalling over roughly $1 trillion over the coming years.

7 QUESTIONS EVERY DATA TEAM SHOULD ASK THE BUSINESS

Most data teams struggle with the same problem: nobody knows what to work on. The business asks for dashboards, the analysts build them, and a month later, everyone quietly stops using them. This piece argues it's a conversation problem. The author's team at Facebook used to sit down with every department twice a year and ask questions that got to the real issues: What's keeping you from hitting your targets? When was the last time data changed your mind? Where are you putting in effort without results? These questions don't sound technical, but they surface where data can actually help. If you stop asking what people “need from data” and start asking how they make decisions, you'll finally find work that matters.

ELAD GIL'S FRAMEWORK FOR SPOTTING BILLION-DOLLAR MARKETS BEFORE THEY LOOK BIG

Markets beat teams beat ideas - great teams can't escape bad markets, but great markets give you multiple shots on goal. Wedding invitations seemed too niche, payroll too boring, file storage too late, then became Minted, Gusto, and Dropbox. Four principles for finding overlooked opportunities: use first principles to ask what's changed in cost or infrastructure, scratch your own itch with real pain, not vague ideas, build product-first because "would you use this" is weak but "would you pay now" is real, and look where smart people tinker for fun before mainstream notices.

Blue Border™ is the #1 VDI alternative

SEMIOTIC

AI-native design agency.

THUNAI

Agentic AI platform that transforms organizational knowledge into self-learning agents for automating support, sales, and marketing tasks.

DATABUTTON

AI-powered dev tool that turns business ideas into full-stack apps through natural language input.

CAN AI WRITE BETTER?

A founder decided to measure what makes great writing tick, literally. By mapping the rhythm and “surprise spikes” in authors like Hemingway and Wodehouse, he found each has a quantifiable fingerprint: where sentences get weird, sharp, or musical. He trained a small model to detect those pulses, and it worked. Maybe the next step in AI isn't more scale, but learning how to imitate the texture of human thought.

BUILD A BILLION-DOLLAR STARTUP WITH A MINIMAL TEAM THANKS TO AI

Instagram sold for $1B with 13 people. WhatsApp had 55. PostHog ships with pods of two to six, and Basecamp runs profitable software with roughly 30 and no managers. The observable pattern is to design for systems over headcount. In practice, that means your engineer answers the support ticket instead of waiting for “customer success,” your designer pushes copy tweaks straight to production instead of filing a Jira ticket, and your founder still writes the onboarding emails because it keeps them close to the user.

CHATGPT FOR COMPETITIVE ANALYSIS

Most people use ChatGPT to ask who their competitors are. This team turned it into a newsroom analyst. They dropped in homepage links, headline lists, and keyword data from Toronto outlets and had it break down tone, coverage gaps, and what kinds of stories each publication actually wins on. It found patterns humans miss, like which sites dominate “how/where/when” searches or bury strong topics under weak headlines. It's the first real glimpse of AI used not for shortcuts, but for curiosity at scale.

COMPOSITE GETS BACKING FROM NFDG FOR ITS CROSS-BROWSER AGENT TOOL

Composite secured $5.6 million in seed funding led by NFDG to develop its cross-browser AI agent tool.

JEFF BEZOS ON WHY WANDERING MATTERS

Jeff Bezos reminds us not to forget the value of “wandering.”

LEARNING TO LET GO: CARE VS. LAISSEZ-FAIRE

VC Gil Dibner reflects on the hardest part of investing: knowing when to step back.

WHY AI AGENTS WILL BE THE DEATH OF THE B PLAYER

The “good enough” era of hiring is ending.

SEE HOW IT WORKS

DATA BREACH AT DOCTORS IMAGING GROUP IMPACTS 171,000 PEOPLE

Doctors Imaging Group, a Florida-based radiology provider, reported a breach where hackers accessed their network for several days in November 2024. Over 171,000 people had personal and medical data compromised, including Social Security and account numbers.

REDIS WARNS OF CRITICAL FLAW IMPACTING THOUSANDS OF INSTANCES

The Redis security team has released patches for a new critical (CVSS 10/10) vulnerability that could allow attackers to gain remote code execution. The vulnerability is caused by a 13-year-old use-after-free vulnerability that can be exploited by authenticated threat actors using a specially crafted Lua script. Redis recommends that administrators patch their servers and, to secure their instances further, they can enable authentication, disable Lua scripting and other unnecessary commands, launch Redis as a non-root user, and enable Redis logging.

THE SALESLOFT-DRIFT BREACH: ANALYZING THE BIGGEST SAAS BREACH OF 2025

The biggest SaaS security incident in 2025 exploited the integration between Salesloft and Drift, compromising over 700 companies. Attackers abused OAuth tokens and SaaS-to-SaaS connections to move laterally, bypassing traditional security defenses and harvesting sensitive credentials. The supply chain nature of the breach allowed a single compromise to spread rapidly across interconnected apps like Salesforce and Gmail, amplifying its impact.

FANTASTIC AWS POLICIES AND WHERE TO FIND THEM

AWS presents a large surface of IAM policies that are available across principal policies, resource policies, AWS Resource Access Manager, resource control policies, and IAM Identity Center permission sets. This post provides a listing and explanation of many of these policies, as well as introducing a new tool called iam-collect. iam-collect can download all IAM policies across every account in an organization for analysis and also provides a set of agent instructions that can be used with an LLM to explore the data further.

FLIPSWITCH: A NOVEL SYSCALL HOOKING TECHNIQUE

FlipSwitch is a novel Linux rootkit technique that bypasses kernel 6.9's new syscall dispatch mechanism by directly patching machine code in the x64_sys_call function rather than overwriting the traditional sys_call_table. The technique locates specific call instructions within the kernel dispatcher and modifies their 4-byte offsets to redirect syscalls to malicious functions, demonstrating how attackers adapt to kernel hardening measures. This research highlights the ongoing arms race between Linux kernel security improvements and rootkit evolution. Elastic has provided YARA signatures to detect this proof-of-concept implementation.

AN IN-DEPTH RESEARCH-BASED WALK-THROUGH OF AN UNINITIALIZED LOCAL VARIABLE STATIC ANALYZER

Cybervelia researchers developed a Binary Ninja engine-based static analyzer that detects uninitialized local variables (ULVs) in stripped binaries by reconstructing stack layouts, tracking read-before-write patterns through Medium-Level IL analysis, and implementing interprocedural taint propagation to trace data flows across function calls. The tool addresses the challenge that modern compiler optimizations obscure ULVs in binaries, despite improved source-level detection, by utilizing techniques such as neighbor-offset heuristics for size inference and proximity-based coverage analysis to identify genuine vulnerabilities. The analyzer integrates IDA Pro symbol importing and Intel PIN execution tracing to reduce false positives and focus on actually executed code paths, demonstrating that ULV detection remains relevant even in modern binaries with recent Linux kernel vulnerabilities serving as examples.

your team. Goodbye firewall alerts, hello smart prioritization.

SETUPHIJACK

SetupHijack is a tool that exploits race conditions and insecure file handling in Windows installer and update processes.

FILIGRAN (PRODUCT LAUNCH)

Filigran unifies threat intelligence, attack surface, and security risk analysis, featuring OpenCTI and OpenBAS for intelligence operations and adversary emulation.

INTRODUCING CODEMENDER: AN AI AGENT FOR CODE SECURITY

CodeMender is a new AI-driven agent developed by Google DeepMind to discover, patch, and prevent software vulnerabilities automatically. The tool addresses threats and proactively secures existing code, enabling developers to build safer software with reduced time spent on manual vulnerability management.

KLOPATRA: EXPOSING A NEW ANDROID BANKING TROJAN OPERATION WITH ROOTS IN TURKEY

Klopatra is a sophisticated new Android banking trojan operated by Turkish-speaking cybercriminals that utilizes commercial-grade Virbox protection and native code libraries to evade detection. The malware targets users in Spain and Italy through fake IPTV apps, exploits Android Accessibility Services for complete device control, and employs both overlay attacks and Hidden VNC capabilities to conduct fraudulent operations while victims are asleep. With over 3,000 compromised devices across two main botnets, Klopatra represents a significant evolution in mobile malware sophistication, demonstrating how threat actors are adopting professional-grade obfuscation techniques typically seen in desktop malware.

‘YOU'LL NEVER NEED TO WORK AGAIN': CRIMINALS OFFER REPORTER MONEY TO HACK BBC

A BBC reporter was contacted via Signal by a “reach out manager” from the Medusa ransomware gang, who offered 25% of the ransom if they gained access to the reporter's corporate device. The reporter played along to gather more information about the gang's tactics, documenting increased pressure from the ransomware actor. Eventually, the actor grew impatient, sending multiple 2FA requests before the reporter could alert the BBC security team, leading to a disconnection from the network.

APPLE PULLS ICEBLOCK FROM THE APP STORE

The ICEBlock app described itself as “Waze but for ICE sightings” and allowed users to report sightings of ICE officials anonymously. AG Pam Bondi stated that Apple removed the app from the App Store after it was requested to be removed because it presents a threat to law enforcement officials, which ICEBlock developer Joshua Aaron has denied, stating it is patently false.

report

GOOGLE'S AI BOUNTY PROGRAM PAYS BUG HUNTERS UP TO $30K

Google has launched a bug bounty program focused on AI security, rewarding up to $30,000 for discovering vulnerabilities that could cause rogue actions or exploit AI systems through issues such as unauthorized account access or data leaks.

DOCKER MAKES HARDENED IMAGES CATALOG AFFORDABLE FOR SMALL BUSINESSES

Docker launched unlimited access to its Hardened Images catalog for small businesses, offering pre-verified container images with near-zero CVEs, up to 95% reduced attack surface, and a 7-day patch SLA through an affordable subscription model.

OPENAI: THREAT ACTORS USE US TO BE EFFICIENT, NOT MAKE NEW TOOLS

OpenAI's October threat report indicates that cyber adversaries primarily leverage AI to enhance their existing hacking techniques, with groups from China, North Korea, and cybercriminal organizations utilizing ChatGPT for malware creation, spear phishing, and reconnaissance activities.

GET THE REPORT →

DUOLINGO ROLLS OUT CHARACTER-LED ADS PLATFORM TO CHARM GEN Z

Duolingo's new advertising product allows brands to run mobile-optimized ads featuring its well-known cast of characters. The company reports that the ads are structured to align with the app's tone and user behavior. The initiative aims to engage and monetize its Gen Z audiences while maintaining user retention and supporting the platform's free education model.

HOW SOCIAL MEDIA GIVES TV A SECOND LIFE

Social media has turned TV shows into cultural ecosystems that live beyond the screen. Gen Z and Millennials now experience TV through TikToks, memes, and online commentary rather than full episodes, making conversation the real content. For broadcasters and brands, this means treating social as a core storytelling format and creating social-first, community-driven narratives that extend a show's impact.

CONTENT AS A SERVICE (CAAS): THE FRAMEWORK THAT ALIGNS MARKETING WITH THE BUSINESS

The “CaaS” framework treats content as a business function rather than a production line. Each asset serves an internal client, such as Sales or Customer Success, with clear goals and metrics tied to business outcomes. Teams prioritize by impact, use intake briefs for alignment, and measure results through data and qualitative proof. The approach replaces busywork with strategy, making content teams accountable for growth instead of output.

THE $5,000 LINKEDIN COMPETITION THAT TURNED OUR TEAM INTO BRAND AMBASSADORS

Storyarb ran a 10-week LinkedIn competition for employees. 14 participants generated over 7M impressions, boosted website traffic by 55%, grew LinkedIn followers by 54%, and influenced sales conversations. The program succeeded by combining financial incentives, gamification, and training on posting skills, hooks, and content strategy. Employees shared posts about topics they were passionate about and turned advocacy into a marketing and recruitment tool.

Plaid case study.

RUNDOWN OF THE "AGENTIC COMMERCE PROTOCOL" FOR CHATGPT'S INSTANT CHECKOUT

The "Agentic Commerce Protocol" enables product eligibility for ChatGPT's Instant Checkout. Similar to Google's product feeds, the protocol allows merchants to provide structured feeds to OpenAI to give more data about their products. There are 9 fields that e-commerce merchants should input and optimize, including title, product_category, popularity_score, and return_rate.

IOS 26 EXPLAINED: WHAT APPLE'S NEW UPDATE MEANS FOR EMAIL & SMS MARKETERS (32 MINUTE VIDEO)

This podcast explains Apple's iOS 26 update and its impact on email, SMS, and web tracking. Link tracking, device signals, and open rates are now less reliable, and SMS messages may land in “Unknown Senders”. It covers attribution challenges, KPI shifts, and multi-channel strategies, while offering tactical fixes like server-side tracking, two-tap opt-ins, and lifecycle flow adjustments to maintain performance.