Tekin Night | Sunday August 9, 2026: CSS Cyber Attacks, Bitcoin Fork & iPhone Ultra

SpaceX Completes Its 50th West Coast Launch

On Thursday morning, August 8th, at 9:35 AM PDT, a Falcon 9 rocket lifted off from Vandenberg Space Force Base in California, carrying 24 Starlink internet satellites to low Earth orbit. This marked SpaceX's 50th launch from the West Coast of America in 2026 - a record demonstrating the company's dominance in the space launch industry.

The booster used in this mission, B1063, traveled to space for the 34th time. This number alone shows how far SpaceX has come in rocket reusability and reuse. Imagine a rocket that has been to space 34 times, is still operational, and can carry heavy payloads to orbit. This is something that seemed impossible years ago.

The Starlink Group 17-38 mission is considered one of SpaceX's routine missions, but the reality is that no launch is "simple." Each launch involves hundreds of checklists, precise preparation, and coordination of multiple teams. After second stage separation and satellite deployment, the first stage booster successfully landed on the droneship "Of Course I Still Love You" - a landing that never stops being amazing.

A Record That Keeps Breaking

SpaceX has become a launch machine in 2026. Fifty launches just from the West Coast means approximately one launch every three days. If we add Florida launches, SpaceX is currently conducting over 120 launches per year - more than all other countries combined.

The Starlink network currently has over 6,000 active satellites in orbit and provides high-speed internet services to more than 4 million users worldwide. With the addition of 24 more satellites, this network continues to expand. SpaceX's ultimate goal is to deploy 42,000 Starlink satellites in multiple orbital layers - a number that's hard to even imagine.

New CSS Attacks: When an Email Can Steal Your Password

At the Black Hat USA 2026 conference held on August 7-8 in Las Vegas, PortSwigger security researcher Gareth Heyes presented several completely new CSS attack methods that can bypass webmail service security boundaries and steal sensitive information. These attacks are not just theoretical - they've been tested in the real world against Gmail, Outlook, Proton Mail, Yahoo, and AOL.

The simplest explanation is this: when you receive an HTML email, your browser or email client executes the CSS code inside it to display the page beautifully. But researchers have discovered that CSS can do much more than beautification - it can steal information, create fake pages, and even manipulate UI behavior.

How Do These Attacks Work?

One of the most interesting attacks presented was the Outlook/Firefox attack chain. In this scenario, an attacker sends an HTML email with special CSS code. When the victim opens this email in Outlook on Firefox, the CSS code forges a Microsoft login page - exactly like the real one. The user thinks their session has expired and enters their password, but they're actually sending it directly to the attacker's server.

Another attack targeted Medium. The researcher showed that CSS attribute selectors can be used to extract Medium authentication tokens that exist in the page DOM. This is done by sending background requests to the attacker's server - something the user doesn't notice.

The more concerning point is that even AI assistants that read and summarize emails (like Gmail's Smart Reply) can be targeted by these attacks. CSS code can show fake content to the AI and cause it to suggest incorrect or dangerous responses.

Webmail Provider Responses

After this research was presented at Black Hat, major webmail providers began investigating and patching these vulnerabilities. Gmail announced it's implementing more restrictions on CSS in HTML emails. Microsoft also said it will soon release an update for Outlook that blocks this attack vector.

But the reality is that these types of attacks have always been a race between attackers and defenders. Today one vulnerability is patched, tomorrow a new method is discovered. The only real way to protect is user awareness and using multiple layers of security.

iPhone Ultra in Two Colors: Silver and Dark Blue

Rumors about Apple's first foldable phone, known as iPhone Ultra, are getting more serious every day. The latest leak from China's supply chain sources shows this device will be offered in two colors: classic silver and dark blue that leans toward ocean depths - not the usual gray seen in Pro models.

This leak comes from photos of iPhone Ultra camera protectors made by accessory suppliers in China. These protectors are typically produced weeks before official unveiling to be ready for initial release. Therefore, the probability of this information being accurate is high.

Apple has positioned iPhone Ultra as a luxury and niche product - not a replacement for iPhone Pro Max, but an entirely new product category. The predicted price is around $2,499, making it the most expensive iPhone in history. At this price, Apple is clearly targeting customers who want to be "first" and are willing to pay for cutting-edge technology.

Why Only Two Colors?

Choosing two colors for the first generation iPhone Ultra makes sense. Apple typically takes a conservative approach with new products and keeps color options limited to make production simpler and quality control easier. Silver is classic and conservative - a color that has always been present in Apple's premium products. Dark blue is bolder and gives those who want to be different an option.

The official unveiling date will likely be at Apple's September event - the same event where the new iPhone line is typically introduced. However, the actual release date might be weeks or even months after the announcement, as mass production of foldable phones is more complex than regular phones.

AMD Acquires Taalas Startup: AI Models Directly in Silicon

AMD announced on August 6th that it has acquired Canadian startup Taalas - a company with a completely unconventional approach to AI chip design. Instead of storing AI models in memory and transferring them repeatedly between processor and memory, Taalas embeds model weights directly into the silicon structure of the chip - meaning the AI model becomes part of the hardware.

This technology is specifically designed for AI Inference - when you use a trained model for prediction and decision-making. Inference is a massive part of the AI market that's growing rapidly, and Nvidia currently has complete dominance. With this acquisition, AMD wants to offer an alternative path.

Direct Attack on Nvidia

This AMD move is a clear message to Nvidia: we're no longer just competing in the general GPU market, but offering specialized solutions for Inference that can be faster and more efficient than you in specific applications. The Inference market is currently worth several billion dollars and is predicted to exceed $50 billion by 2030.

Taalas is based in Toronto, Canada, and its team includes researchers and engineers who have worked for years on hardware architectures for machine learning. AMD hasn't disclosed the purchase price, but given the strategic importance of this technology, a significant amount was likely paid.

Progress Kemp LoadMaster on CISA's Dangerous List

On August 7th, the U.S. Cybersecurity and Infrastructure Security Agency (CISA) added vulnerability CVE-2026-8037 in Progress Kemp LoadMaster to the KEV (Known Exploited Vulnerabilities) catalog. This means one thing: this vulnerability is currently being exploited in the real world and is considered an immediate threat.

CVE-2026-8037 is a Command Injection flaw with a critical CVSS score of 9.6 out of 10. This vulnerability allows an unauthorized attacker to execute arbitrary system commands on the LoadMaster appliance. LoadMaster is a load balancer typically placed at the network edge that terminates HTTPS traffic - meaning it decrypts it before reaching internal servers.

Active Exploitation Attempts

According to reports, 792 exploitation attempts of this vulnerability have been recorded so far. This number shows that attackers are actively scanning the internet to find vulnerable devices. Progress, the LoadMaster manufacturer, has released a patch that fixes this issue, but unfortunately many organizations haven't updated yet.

CISA has ordered all federal agencies to patch this vulnerability by August 28th or remove vulnerable devices from the network. This is a firm order indicating the seriousness of the threat. Private sector organizations are also strongly recommended to do the same.

Bitcoin Fork: When the Chain Splits at Block 961,632

On August 8th, 2026, something happened that is rare but not unprecedented in Bitcoin history: the Bitcoin chain split into two parts at block 961,632. The reason? A fundamental disagreement about BIP-110 - a proposal to restrict arbitrary data in Bitcoin transactions.

BIP-110 proposes that Bitcoin should enforce stricter rules on the type of data that can be included in transactions. The goal is to prevent excessive use of Bitcoin for storing NFTs, images, and non-financial data. But this proposal is controversial - some believe Bitcoin should be free and anyone should be able to store anything in it, while others say it should only be used for financial transactions.

What Happened?

At block 961,632, the Roughnecks mining pool decided to activate BIP-110 rules and mine blocks that only accept transactions compliant with BIP-110. This caused a separate chain to be created, as other nodes that hadn't accepted BIP-110 rejected these blocks and continued the main chain.

Roughnecks has mined two blocks on the BIP-110 chain so far, but the main chain is 18 blocks ahead. This is a significant difference and shows that the vast majority of the Bitcoin network doesn't support BIP-110. Statistics show that only 2.6% of the network's mining power agreed with BIP-110 - far less than the 55% threshold needed for a successful fork.

What's the Future of This Fork?

Given the lack of support, there's a high probability that the BIP-110 chain will soon be abandoned and miners will return to the main chain. In Bitcoin history, forks with less than 10% support have typically had short lifespans. For a fork to survive, it needs economic support, exchanges that accept it, and wallets that support it.

Current Bitcoin mining difficulty is at 127.48 trillion - a new record showing the network's very high security. With this difficulty, small forks practically have no chance to compete with the main chain unless they attract significant support.

Tekin Analysis: Sunday Night and the Future of Technology

Three Key Lessons from Tonight's News

First, reuse and recycling is now the standard of the space industry. A booster that's been used 34 times shows that SpaceX is no longer in the testing phase - it's in the industrial operations phase. This is something that seemed like fantasy 10 years ago.

Second, cybersecurity is an endless race. Today CSS attacks, tomorrow something else. Webmail providers must always stay one step ahead, and users must always be vigilant. There's no permanent solution - only continuous improvement.

Third, technology is always political and philosophical too, not just technical. The Bitcoin Fork isn't about code - it's about what Bitcoin should be. A pure payment system? Or a public data platform? These are philosophical questions that can't be solved with code.

Conclusion: A Sunday Night Full of Transformation