October 31, 2026

BerLabs — October 31, 2026

Binance hackers stole $6.3 billion in record-breaking crypto heist

Binance disclosed a $6.3 billion crypto theft, The Verge reports. The exchange says about $3 billion in affected assets have been frozen and promises to cover losses with company funds. Investigators initially identified a compromised engineer’s laptop, but Binance says additional breaches were involved; TRM and Chainalysis attribute the attack to North Korea’s TraderTraitor group.

Why it matters
The suspended transactions and enormous reported loss make this a major test of exchange security, incident response, and customer protection.

Read the source →

2027 Nissan Rogue Hybrid: Drives like an EV, but there's no plug involved

Nissan is bringing its latest e-Power system to the 2027 Rogue Hybrid, Ars Technica reports. Its gasoline engine charges the traction battery rather than driving the wheels through a transmission. That lets Nissan optimize the engine for generating electricity, offering a concrete look at how a series hybrid differs from more familiar parallel designs.

Why it matters
The design gives buyers another route to electric propulsion without plugging in, though the supplied excerpt lacks fuel-economy results.

Read the source →

The touchscreen MacBook Pro is just a MacBook Pro with a touchscreen

The Verge’s hands-on finds the new touchscreen MacBook Pro familiar: taps and swipes largely duplicate trackpad functions, with some interface elements expanding when touched. The reviewer found the OLED display and thinner, lighter chassis more striking than touch itself. Professionals at Apple’s briefings still defaulted to keyboards and trackpads, treating the screen as an extra input option.

Why it matters
Mac users gain another way to interact with their laptops, while the initial experience suggests little need to relearn existing workflows.

Read the source →

Catching Bugs Before They Land: JiTTesting’s Human-on-the-Loop Approach at Full Scale

Meta says its deployed JiTTests system generates tests that reveal behavioral differences in risky code changes, then filters findings with rules and multiple language models. Engineers assess whether flagged changes were intended. Meta reports a 60% true-positive rate and says the system helped catch 74 serious privacy bugs before production in the first half of 2026.

Why it matters
Executable failures give reviewers concrete evidence to investigate, offering a practical complement to AI-generated code-review comments.

Read the source →

Cloudflare DDoS Threat Report 2026 Q3: More swarms, fewer giants

Cloudflare says it mitigated 18.2 million network-layer DDoS attacks in Q3, up 208% from a year earlier. Across its network, 96% ended within 10 minutes and 98% stayed below 1 Gbps. Attacks exceeding 1 Tbps fell from 805 in Q2 to 251, while DNS amplification increased sharply.

Why it matters
Frequent, brief attacks leave little time for manual response, making automatic mitigation relevant even when headline-sized floods become less common.

Read the source →

Octoverse: Commits, pushes, reach new highs in 2026 as developers build with agents

GitHub’s Octoverse report puts agent-started pull requests above 40% in September, up from 17% in March. The platform recorded 17.9 billion code pushes over the past 12 months, exceeding 2024 and 2025 combined. New repositories with agent configuration files averaged eight times as many commits as those without, an association that does not establish better software or higher productivity.

Why it matters
The figures show automated development becoming routine on GitHub, increasing the importance of review as code activity grows.

Read the source →

123 Reg discloses 'AI cyberattack' on customer data, suspects China nexus

Domain registrar 123 Reg says a September 18 intrusion exposed customer names, phone numbers, VAT numbers, preferences, and account product information, The Register reports. The company says attackers may have copied the data and is notifying customers and regulators. It describes the attack as AI-assisted and suspects a China-linked actor, but those conclusions remain its assessment.

Why it matters
Exposed account details could make impersonation attempts more convincing for affected customers.

Read the source →

API credits benefit for eligible ChatGPT subscriptions

OpenAI has added API credits equal to the subscription price for eligible $100, $200, and $500 plans, according to Simon Willison. The credits can be applied to one API organization. Willison compares the benefit with Anthropic’s recently announced offer and says it could encourage him to publish demos using credits already covered by his subscriptions.

Why it matters
Developers already paying for eligible plans may have API capacity available for experiments without a separate initial outlay.

Read the source →