60
Everything in Moderation: Per-Domain Coverage Optima and Alignment-Resistant Domain Gaps in Multi-Domain Mid-Training
Controlled mid-training experiments on Qwen3-8B-Base find each domain has a 10-40% coverage optimum and domain gaps survive alignment SFT.
Using Qwen3-8B-Base (with a 4B replication) across five semantically rule-disjoint KOR-Bench domains, the authors train 30 data allocations spanning the five-domain simplex at five seeds each. All five domains show interior optima in the moderate 10-40% coverage band, and domain gaps persist after a fixed-budget compensatory SFT pass, which raises 116/120 cells yet bridges 0/240 pairs at a 5% threshold. Zero coverage collapses mid-training-only accuracy, though a FineWeb-Edu-only control shows the collapse is partly generic drift. The results argue mid-training data composition requires principled design rather than reliance on later alignment.
30
Unit 42 Researchers Discover Multiple Espionage Operations Targeting Southeast Asian Government
Unit 42 attributes three espionage clusters targeting a Southeast Asian government to Stately Taurus, Alloy Taurus, and Gelsemium APTs.
Unit 42 investigated espionage attacks starting in late 2022 against multiple governmental entities in a Southeast Asian country, including critical infrastructure, public healthcare, financial administrators, and ministries. Analysis revealed three distinct clusters: CL-STA-0044 attributed to Stately Taurus (Mustang Panda), CL-STA-0045 to Alloy Taurus (GALLIUM), and CL-STA-0046 to Gelsemium. The first cluster used a ToneShell backdoor variant, ShadowPad, China Chopper web shells, Impacket, and credential dumping tools across roughly Q1 2021 to Q3 2023. All three operated with distinct tools, infrastructure, and long-term surveillance tradecraft consistent with APTs.
70
30
30
30
30
30
42
47
30
55
Inside ‘Project Lily’: The Humans Reading Your ChatGPT Chats
404 Media reveals OpenAI's 'Project Lily' has hundreds of contractors reading real ChatGPT user prompts, exposing sensitive personal data despite privacy filters.
404 Media reports that OpenAI employs hundreds of contractors who read real ChatGPT user prompts, including whole conversations, to rate and critique the chatbot's responses across a user base of over 900 million. Prompts are anonymized and run through OpenAI's Privacy Filter model, but the company acknowledged sensitive personal details can still reach reviewers, and 'user memories summaries' may reveal a user's location and personal context. The review work includes training ChatGPT to be less sycophantic and to stop anthropomorphizing itself, following lawsuits linking the sycophantic 4o model to multiple suicides. Anthropic confirmed it also uses human review to improve its models, and OpenAI's 'improve the model for everyone' data-sharing setting is on by default for free, Plus, and Pro users.
55
Pornhub's Parent Company to Pay $120 Million to Settle Child Sexual Abuse Lawsuits
Pornhub parent Aylo will pay $120 million settling child sexual abuse class actions, without admitting liability, and adopt stricter content moderation commitments.
Aylo, Pornhub's parent company (formerly Mindgeek, acquired by Ethical Capital Partners in 2023), will pay $120 million to settle two 2021 class actions in California and Alabama alleging its platforms hosted child sexual abuse material in violation of federal trafficking and child imagery laws. The settlement fund begins with $25 million in 2026 followed by six annual installments. The class covers anyone under 18 appearing in content on Mindgeek-operated sites between February 12, 2011 and December 6, 2024. The deal, subject to court approval, adds commitments to age-verify models, conduct human and automated content review, and report suspected abuse material to authorities.
20
55
How Tea’s Founder Convinced Millions of Women to Spill Their Secrets, Then Exposed Them to the World
42
45
45
45
30
30
45
57
30
57
Persistent Attempts at Cyberespionage Against Southeast Asian Government Target Have Links to Alloy Taurus
Alloy Taurus (GALLIUM) compromised Southeast Asian government networks from 2022 to 2023 using Exchange web shells and undocumented .NET backdoors Reshell and Zapoa.
Unit 42 tracked persistent multiwave intrusions at a Southeast Asian government starting in early 2022 and continuing through 2023, attributing the activity with moderate confidence to Alloy Taurus (aka GALLIUM), a Chinese state-aligned espionage group. Attackers exploited Exchange Server vulnerabilities to deploy web shells including China Chopper, then ran reconnaissance with Fscan and WebScan, created administrative accounts, and installed undocumented .NET backdoors named Reshell and Zapoa. They established resilience by installing SoftEther VPN, brute-forced Active Directory credentials with Kerbrute, and dumped credential stores with GoDumpLsass and LsassUnhooker. The campaign reflects long-term espionage tradecraft to maintain a foothold.
70
55
30
30
30
30
30
55
45
42
30
55
57
30
45