Author: Carter

Charles Hoskinson is building a more compelling argument for artificial intelligence as development on Midnight City advances. The Cardano founder now views AI agents as the foundation of how the network communicates and grows. Here’s a summary of his recent comments, along with a deeper look at what Midnight City is aiming to demonstrate. How Hoskinson Views AI’s Role in Cardano’s Next Chapter AI agents operate on their own, handling trades, posting content, and coordinating tasks without a person at the controls. Hoskinson has embraced this concept after facing criticism over recent experiments connected to Cardano’s official social channels. The…

From cloud computing to worldwide food supply chains, no industry is safe from the constantly changing nature of cyber threats. The 2025 sector cyber threat reports recently published by IT-ISAC and Food and Ag-ISAC highlight a clear truth: organizations in every field are up against determined adversaries, including both state-backed hacking groups and organized cybercriminal networks. These reports also shed light on how adversaries are shifting their techniques and how rising global tensions are making the threat landscape more complicated for all organizations, no matter the industry. Understanding Risk Through the PASS Framework To get a clearer picture of the…

In my previous post, I discussed how to obtain structured, machine-readable outputs from an LLM using JSON Mode, function calling, and structured outputs. In that discussion, we briefly explored function calling as a way to retrieve structured responses. However, function calling is much more than just a way to get organized data back from a model — it is, in fact, the foundation of agentic AI workflows. So, in this post, we’re going to dive deeper into this subject. In every example we’ve discussed up to now, the LLM has been treated merely as a passive respondent — it receives…

I recently revisited my matrix recurrent units algorithm (the MRU), a novel linear-time sequence architecture I created as an alternative to attention. I explain it in depth at the repo, but the gist is the MRU works by transforming the embedding into an input state matrix, cumulatively multiplying the matrices across the sequence dimension to get the output state matrix, and then transforming the matrices back into a vector. In order to make the MRU efficient on DL hardware, I created a parallel scan by utilizing the operation's associativity. About a year ago, I shared my project on Reddit (I've…

In brief Mercury 2, developed by Inception Labs, produces around 1,000 tokens per second and earned a score of 90 on the AIME 2026 benchmark. Google’s DiffusionGemma reaches comparable speeds but falls short on performance benchmarks. DiffusionGemma is freely available as an open-weight model on Hugging Face, whereas Mercury 2 is a closed-weight, paid API offering. On Thursday, Inception Labs unveiled Mercury 2, branding it as the planet’s quickest reasoning language model. According to the company’s release, it churns out approximately 1,000 tokens per second—the fragments of text an AI system processes and generates—compared to roughly 89 tokens per second…

`Ravie Lakshmanan`Jun 19, 2026Threat Intelligence / Firewall Security On Thursday, the U.S. Cybersecurity and Infrastructure Security Agency (CISA) advised Fortinet clients using FortiGate devices to implement security measures. This comes in response to a large-scale attack targeting thousands of devices accessible over the internet. Known by the codename FortiBleed, this extensive campaign is being linked to Russian-speaking hackers. As of June 19, 2026, reports indicate that 86,644 devices have been compromised. SOCRadar’s data shows that a majority of the stolen credentials are generic admin accounts (35%) and built-in Fortinet system accounts (28.3%). Specific organizational accounts make up the remaining 36.7%…

When NASA launched the Artemis II space vehicle in early April, a 25-year-old software framework was quietly powering many of its critical systems. But don’t mistake this for some outdated relic that the space agency simply forgot to replace. On the contrary, Artemis II leaned on the core Flight System (cFS) to manage everything from telescope operations and avionics to command and control — because software engineers have been steadily refining and improving cFS since the early 2000s. NASA develops and sustains cFS through an open source platform and counts on the wider space community to keep the architecture, tools…

Kyle Kucharski/ZDNETFollow ZDNET: Add us as a preferred source on Google.ZDNET’s key takeaways The permissions you grant smartphone apps can quietly compromise your privacy.Periodically reviewing app permissions helps safeguard your personal data from unnecessary exposure.Start by auditing the most sensitive permissions, then make it a habit to revisit them on a regular basis.Your smartphone, regardless of whether you prefer Android, iOS, or some other niche mobile operating system, can leave digital trails that anyone with the know-how can follow.Every app I install demands certain permissions to function. If you want food delivered, the app may need GPS access to locate…

Nous Research has added a Blank Slate setup mode to its open-source Hermes Agent. It inverts the usual onboarding. Instead of a fully loaded default, you start with almost nothing. Hermes Agent is the self-improving agent framework from Nous Research. It runs on your own machine. The team announced the new mode on X. Blank Slate now joins two existing options: Quick Setup and Full Setup. TL;DR Blank Slate boots an agent with everything off except provider & model, File Operations, and Terminal. Web, browser, code execution, vision, memory, delegation, cron, skills, plugins, and MCP stay disabled. It writes an…

def make_rag_chunks(rows, max_chars=700): chunks = [] for row in rows: text = ( row.get(“text_preview”) or row.get(“rendered_text”) or row.get(“description”) or “” ) text = normalize_text(text) if not text: continue sentences = re.split(r”(?<=[.!?])s+”, text) current = “” for sentence in sentences: if len(current) + len(sentence) + 1 <= max_chars: current = (current + ” ” + sentence).strip() else: if current: chunks.append( { “chunk_id”: hashlib.sha1( (row.get(“url”, “”) + current).encode() ).hexdigest()[:12], “url”: row.get(“url”), “source”: row.get(“source”), “page_type”: row.get(“page_type”), “title”: row.get(“title”) or row.get(“name”), “text”: current, } ) current = sentence if current: chunks.append( { “chunk_id”: hashlib.sha1( (row.get(“url”, “”) + current).encode() ).hexdigest()[:12], “url”: row.get(“url”), “source”: row.get(“source”), “page_type”:…