Real-time Controversies
How the Real-time Controversies feature (part of the Risk module group) reads global news, identifies genuine ESG controversies, groups coverage into incidents, and scores each one.
Real-time Controversies reads global news daily, finds genuine ESG controversies involving specific companies, groups all the coverage of each one into a single incident, and gives each incident a severity score. It is part of the Risk module group in the Compliance and Reporting product.
For the scoring methodology — the four dimensions, the formula, and a worked example — see How controversy scores are calculated.
Coverage and freshness
- Around 35 trusted sources. A vetted list of outlets chosen for credibility — global wire and broadcast services (such as BBC, AP, The Guardian and CNN), Nordic, German-speaking and Benelux public broadcasters (such as SVT, NRK, YLE and Tagesschau), other regional public broadcasters, and ESG trade outlets (such as ESG Today and Carbon Brief). Keeping the list curated is a quality control on sources, not a way of chasing volume — it reduces noise from outlets republishing each other's stories. Sources can be added on request.
- Five languages read directly — English, Swedish, Finnish, Norwegian and German — with summaries and titles delivered in English.
- 24 ESG themes — eight environmental, eight social, eight governance (listed below).
- Daily updates, covering news up to the most recent full day.
- Company matching by identity, not name. Each company is matched using a cross-referenced identity rather than a plain name search, then disambiguated — so companies with similar names are not confused.
How an article becomes an incident
Seven steps take a raw news article to a scored controversy. Everything after collection is done by AI models, with each incident's reasoning recorded along the way.
- Collect — news for each monitored company, in five languages
- Clean — remove audio items and obvious duplicates
- Filter — keep only genuine, negative, company-specific controversies
- Theme — tag against the 24 ESG themes and extract the key facts
- Group — attach every article about the same event to one incident
- Score — rate each incident across four dimensions
- Deliver — flag repeated wire copy and output the records
Two of these steps do the most to keep the data clean:
Filtering — telling a real controversy from noise
An article only counts when the company itself is accused of, or directly responsible for, the conduct — and the underlying event is negative. Earnings coverage, good-news PR, sector-wide reports with no company-specific finding, and one-off personal misconduct by an employee are all removed before anything is themed or scored.
Grouping — one incident, not fifty headlines
Articles are matched to incidents by their meaning, not just their wording. A follow-up story, a company statement, or a later fine attaches to the existing incident rather than creating a new one. The result is clean incident counts with no double-counting — an ongoing legal case is one incident, however many articles cover it.
The 24 ESG themes
Every incident is tagged with one or more themes, ordered by relevance.
| Environmental | Social | Governance |
|---|---|---|
| Climate and emissions | Child labour | Corruption and bribery |
| Pollution and waste | Forced labour and trafficking | Financial misconduct |
| Biodiversity and land use | Wages and benefits | Anti-trust |
| Resource depletion | Health and safety | Executive pay and board governance |
| Water | Diversity, harassment and discrimination | Tax transparency |
| Animal welfare | Community and indigenous rights | Cybersecurity |
| Greenwashing | Product safety | Controversial weapons |
| Other environmental issues | Data privacy and AI ethics | Other governance issues |
What an incident record contains
Each incident is a single structured, machine-readable record:
- The company involved and the ESG theme or themes it relates to
- Where the harm sits — own operations, supply chain, product or service, or investment
- The four dimension scores, each with written reasoning, plus the combined controversy score, the four-tier severity label, and a confidence rating
- A short incident summary, the core issues, and what is new since the last update
- Every source article, with duplicates flagged and the main source identified
- The dates first seen and last updated, and a stable incident ID that persists as the story develops
How it appears in the product
This section will describe how Real-time Controversies surfaces in the Compliance and Reporting product — alerts, incident views, and severity tracked over time. The underlying data is also suitable for delivery as a data feed.
What Real-time Controversies does not cover
Real-time Controversies is news-driven: it tracks incidents as they are reported. It is distinct from Exclusion Screening, which flags or excludes holdings against configurable policies covering business involvements, norms compliance, sanctioned nations, and public exclusion lists. The two are complementary — a controversy may inform an exclusion decision, but the features answer different questions.