GoKawiil
Tech News
New York City has imposed a temporary ban on artificial intelligence use in classrooms through middle school, while also setting new limits on student screen time. High school students, by contrast, will gain access to AI literacy courses under the new policy.
A reply-post analyzing a reader's use-after-free bug in a limit-order matching engine argues that object pools change the nature of memory bugs rather than eliminating them. The author explains that without pooling, a freed-and-reused memory slot can let two different object types collide, potentially turning a stale pointer into a code-execution exploit via a mismatched function pointer. With pooling, the same logical use-after-free persists, but because the recycled memory always holds the same type, the dangerous type confusion is avoided—unless the object itself contains an inline tagged union, which reopens the original hazard.
Pangram co-founder and CEO Max Spero appeared on TechCrunch's Equity podcast to discuss the complexities of detecting AI-generated content. The startup recently raised $9 million, partnered with Substack to flag AI-written newsletters, and launched a new tool for detecting AI-generated images.
X announced that as of September 2, 2026, all U.S. creator payouts—including Original Content Rewards and subscription earnings—will be processed through its own X Money platform rather than Stripe. The new system sends funds instantly rather than on the previous biweekly schedule with a $30 minimum payout. The company's messaging suggests creators no longer have an alternative payment option, effectively mandating the switch.
A federal judge who previously found Google operates an illegal monopoly in advertising technology has stopped short of ordering the company to sell off that business. Instead, the ruling leaves Google's ad tech operations largely intact, drawing sharp criticism from advocacy groups who argue the remedy fails to match the severity of the monopoly finding.
Florida International University researchers tested the Apple Watch Series 8, Fitbit Sense 2, Samsung Galaxy Watch 5 and Garmin Forerunner 955 against a Cosmed K5 metabolic device and ECG chest strap during controlled workouts with 58 participants. All four smartwatches overstated calorie burn, with margins of error ranging from 15% to 25%, meaning a reported 200-calorie burn could really be closer to 160-174 calories. Apple Watch was the most accurate, while Samsung and Garmin overestimated more, and Fitbit had enough tracking glitches that some of its data had to be excluded.
The Trump administration filed a 20-page brief supporting OpenAI in its copyright dispute with The New York Times, arguing that training large language models on copyrighted material qualifies as fair use. The filing cites an executive order calling for the US to maintain global leadership in artificial intelligence, and warns that restricting AI training practices could harm American innovation and economic competitiveness.
Nexus Mods, the largest PC mod-sharing platform, has purchased SteamDB, the long-running independent site that tracks Steam pricing, updates and player data. The deal followed talks with xPaw, the sole maintainer of SteamDB, who was seeking more resources and stability than he could provide on his own. SteamDB will keep its name and current look for now, with Nexus Mods handling infrastructure and funding behind the scenes.
An independent developer ran a three-week ChatGPT ad campaign for event-seating software PerfectTablePlan, spending £289.52 for 2,988 clicks at an average 10p per click. The campaign yielded just 14 installs, a 0.46% conversion rate and a 69% bounce rate, versus roughly 5% conversion and under half the bounce rate on Google Ads and organic ChatGPT referrals. Working with ad-fraud researcher Dr Augustine Fou, the developer used analytics tooling to rule out bot traffic, pointing instead to poor targeting as the likely cause of the weak results.
Federal Judge Leonie Brinkema rejected the DOJ's push to force Google to divest its AdX ad exchange, despite having ruled last year that Google illegally monopolized two ad tech markets. Instead, she accepted unspecified 'behavioral remedies' that reportedly require Google to open its ad tech tools to competitors, with the ruling issued under seal for now.
Security firm Horizon3 says hackers are exploiting CVE-2026-9586, an unauthenticated SQL injection bug in Sangoma Switchvox's /pa endpoint that allows remote code execution. Honeypot data shows a single source IP rapidly hitting multiple exposed systems, deploying reverse shells and exfiltrating process data. Sangoma patched the flaw, one of 12 reported by Horizon3, in version 8.4.0.2 released July 14.
An entrepreneur used AI extensively to make pricing, hiring and marketing decisions while scaling a services business, reaching her $50,000 monthly revenue goal within six months. However, her underpriced premium offerings meant that payroll, rent and overhead grew alongside revenue, leaving no profit to reinvest or pay herself. The strain eventually caused missed deadlines, declining quality and staff departures.